Directory UMM :Data Elmu:jurnal:B:Biological Psichatry:Vol48.Issue6.2000:

The Impact of Substance Abuse on the Course of
Bipolar Disorder
Stephen M. Strakowski, Melissa P. DelBello, David E. Fleck, and Stephan Arndt
Background: Substance abuse occurs at high rates in
bipolar disorder. The reasons for this co-occurrence are
unknown. Alcohol use disorders have been associated with
both earlier and later age of onset of bipolar disorder, in
part based on the temporal associations of the two
conditions. Both drug and alcohol use disorders are
associated with impaired outcome of bipolar illness. This
influence may involve both direct effects of alcohol or
drugs on the initiation of affective symptoms and indirect
effects on treatment compliance. To extend these previous
findings we examined the temporal associations of substance abuse and affective symptoms in patients with new
onset bipolar disorder.
Methods: Associations between affective symptoms and
alcohol and cannabis use disorder symptoms were evaluated using regression and time-series correlative methods
in 50 new-onset bipolar patients.
Results: The duration of alcohol abuse during follow-up
was associated with the time patients experienced depression. The duration of cannabis abuse was associated with
the duration of mania. Several subgroups could be identified with different temporal relationships among these

disorders.
Conclusions: Although the relationships among substance
use and bipolar disorders are complex, systematic study of
the courses of the disorders might clarify how these
conditions interact longitudinally. As the numbers of
subjects in specific subgroups are relatively small in this
study, these results should be considered preliminary.
Biol Psychiatry 2000;48:477– 485 © 2000 Society of Biological Psychiatry
Key Words: Bipolar disorder, alcohol abuse, cannabis
abuse, outcome, review

From the Bipolar and Psychotic Disorders Research Program, Department of
Psychiatry, University of Cincinnati College of Medicine, Cincinnati, Ohio
(SMS, MPD, DEF) and the Department of Psychiatry, University of Iowa, Iowa
City (SA).
Address reprint requests to Stephen M. Strakowski, M.D., University of Cincinnati
College of Medicine, Bipolar and Psychotic Disorders Research Program,
Department of Psychiatry, 231 Bethesda Ave, PO Box 670559, Cincinnati OH
45267-0559.
Received January 8, 2000; revised March 30, 2000; accepted April 4, 2000.

© 2000 Society of Biological Psychiatry

Introduction
The Prevalence of Substance Abuse in Bipolar
Disorder

S

ubstance abuse is exceptionally common during the
course of bipolar disorder. For example, the National
Institute of Mental Health (NIMH) Epidemiologic Catchment Area (ECA) study found that over 60% of patients
with type-I bipolar disorder developed a substance use
disorder during their lifetime (Regier et al 1990). In fact,
in that study, co-occurring alcohol or drug abuse or
dependence was more common in bipolar disorder than
any other Axis I syndrome. These elevated rates of
substance use disorders do not simply reflect the effects of
chronic mental illness, since rates of substance use disorders are elevated in bipolar patients even at the time of
their first manic episode (Strakowski et al 1993, 1998a,b).

Moreover, high rates of substance abuse in bipolar disorder have been reported from many different countries,
treatment settings, and clinical and research samples
(Fogarty et al 1994; Strakowski et al 1994; Verdoux et al
1996; Winokur et al 1995). Recent reports from Taiwan,
however, in which the lifetime rate of substance abuse in
bipolar patients appeared to be quite low (,10%; Tsai et
al 1996; Tsai et al, in press), suggest that this cooccurrence may be primarily restricted to Western societies. Additional reports from other non-Western countries
are needed to replicate this observation.
The reasons substance abuse is so common in bipolar
patients are unknown (Strakowski and DelBello 2000).
One explanation is that patients with both bipolar and
substance use disorders are more likely to seek psychiatric
treatment than patients with only one of the two conditions
(the so-called “Berkson’s bias”; Berkson 1946). This
explanation fails, however, because studies using population survey methods found elevated rates of substance use
disorders in bipolar disorder that were as high or higher
than those reported in clinical samples (Kessler et al 1997;
Regier et al 1990). Alternatively, because one of the
criteria for mania is an “excessive involvement in pleasurable activities that have a high potential for painful
consequences” (American Psychiatric Association 1994),

which could be interpreted to include substance abuse,
0006-3223/00/$20.00
PII S0006-3223(00)00900-8

478

BIOL PSYCHIATRY
2000;48:477– 485

increased rates of substance abuse in bipolar disorder
might result simply by definition. In our previous work,
however, substance abuse alone was not considered adequate to meet this criterion, and yet rates of substance use
disorders were elevated in our bipolar samples (Strakowski et al 1993, 1996a, 1998a, 1998b). Moreover,
unlike the general population, bipolar patients are more
likely to exhibit substance dependence than substance
abuse (Regier et al 1990). A diagnosis of substance
dependence requires evidence of sustained drug and alcohol misuse that imparts significant social and functional
consequences. A bipolar patient who abuses drugs or
alcohol simply as part of the pleasure seeking of mania is
unlikely to exhibit these longitudinal characteristics.

Therefore, the high rate of substance use disorders in
patients with bipolar disorder is not explained by ascertainment or diagnostic bias. Recently, we and others
suggested that the co-occurrence of bipolar and substance
use disorders results from a combination of patients with
different types of associations among these two conditions
(DelBello and Strakowski, in press; Mueser et al 1998;
Strakowski and DelBello 2000). Specifically, one subgroup of patients includes subjects in whom bipolar illness
initiates substance abuse, either as an attempt at selfmedication or as a direct result of affective symptoms
(e.g., the increased pleasure seeking of mania); however,
evidence supporting this relationship between bipolar and
substance use disorders is sparse (DelBello and Strakowski, in press; Strakowski and DelBello 2000), suggesting this subgroup is only a small fraction of the total
patients. A second subgroup includes subjects in whom
substance abuse has initiated affective episodes, perhaps
through behavioral sensitization or kindling mechanisms
(Sonne et al 1994). Clinical data suggest that this subgroup
may be relatively common, as many patients have the
onset of substance abuse prior to the onset of affective
illness (Strakowski and DelBello 2000; Strakowski et al
1996a) and bipolar patients with co-occurring substance
abuse may have lower familial rates of affective illness

(DelBello et al 1999) suggesting an additional risk factor
(i.e., substance abuse) is needed to precipitate bipolar
disorder in those individuals (Winokur et al 1995). A third
subgroup consists of subjects who share a common risk
factor for both disorders. One such risk factor could be a
gene or genes associated with both disorders; however,
family studies have not supported this suggestion (reviewed in Strakowski and DelBello 2000). A shared
vulnerability toward behavioral sensitization or kindling,
which might underlie both disorders, can also be hypothesized, although this has not been studied in any detail
(Post 1992; Strakowski et al 1996b). Alternatively, psychosocial stress might precipitate both conditions (reviewed in Strakowski and DelBello 2000). Finally, be-

S.M. Strakowski et al

cause substance use disorders are common in the general
population, then a portion of the bipolar patients would be
expected to develop substance abuse or dependence simply by chance. Unfortunately few published studies have
specifically attempted to identify subgroups of patients
with co-occurring bipolar and substance use disorders to
clarify these relationships.

Substance Abuse and the Age of Onset of Bipolar
Disorder
Although the specific reasons for the common co-occurrence of bipolar and substance use disorders has remained
elusive, nonetheless, substance abuse appears to influence
the age of onset and subsequent course and outcome of
bipolar disorder. In order for substance abuse to impact the
age of onset of bipolar disorder, it must begin prior to the
onset of affective symptoms. Previous studies suggest that
this occurs in approximately 60% of patients with both
conditions (reviewed in Strakowski and DelBello 2000).
In one of the earliest studies to examine whether alcohol
abuse influenced the age of bipolar disorder onset, Morrison (1974) reported that affective illness was first identified at age 23 years in nonalcoholic bipolar patients but
not until age 28 in alcoholic bipolar patients. Similarly,
Strakowski and colleagues (Strakowski et al 1996a,
1998b) found that patients hospitalized with first-episode
mania and an antecedent alcohol use disorder (i.e., beginning at least 1 year prior to the affective illness) had a
significantly later age of onset (mean of 27 years) than
those without alcoholism (mean of 21 years). In contrast,
Winokur et al (1996) found that bipolar patients with
alcoholism had a significantly younger age of onset (mean

of 23 years) than those without (mean of 27 years). These
conflicting results are difficult to reconcile. One explanation was suggested in a separate study by Winokur et al
(1995). They found that bipolar patients with alcoholism
that predated the affective illness (i.e., antecedent alcoholism) had a later age of onset (27 years) than those patients
whose alcoholism began after the onset of bipolar disorder
(20 years). They suggested that the early onset group had
a more severe bipolar illness that included the development of alcohol abuse, whereas the patients with antecedent alcohol abuse had a less severe illness that required the
presence of alcoholism to initiate the bipolar disorder. By
identifying specific subgroups (e.g., early onset bipolar
patients with antecedent alcohol abuse) it may be possible
to better define these relationships among alcohol use and
bipolar disorders.
There are few studies of associations between drug use
disorders (other than alcohol) and the age of bipolar
disorder onset. Strakowski et al (1996a) did not find any
difference in the age of bipolar onset in patients with or

Substance Abuse in Bipolar Disorder

without antecedent drug (mostly cannabis) abuse or dependence; however, Sonne et al (1994) found patients with

substance abuse had an earlier age of onset of bipolar
illness, but they combined drug and alcohol abuse into a
single category so that the effects of specific substances
could not be determined. The converse relationship of how
bipolar disorder might influence the onset of substance
abuse has not been studied systematically to our knowledge. The study by Winokur et al (1995), however,
suggests that early onset bipolar disorder may place
patients at risk for developing substance abuse.

Substance Abuse Impacts Outcome of Bipolar
Disorder
A number of investigators reported that substance abuse
has negative effects on outcome in bipolar patients. In
1974, Reich and colleagues observed that patients with
bipolar disorder who required hospitalization were significantly more likely to have a history of excessive alcohol
use compared to patients who did not require hospitalization (Reich et al 1974). Shortly thereafter, Himmelhoch et
al (1976) observed that alcohol and drug abuse was
associated with mixed states that are associated with poor
outcome themselves. In a regression analysis, they noted
that both mixed states and drug abuse were associated with

poor treatment outcome, but that the effects of drug abuse
were “more profound” than those of the mixed symptoms.
In a 4-year prospective outcome study of 75 bipolar
patients following hospital discharge, Tohen et al (1990)
examined a number of variables in a Cox regression model
to identify independent predictors of outcome. A history of
alcoholism was associated with poor occupational status at
6 and 48 months after hospital discharge, even after
controlling for a number of other variables. Feinman and
Dunner (1996) noted that substance abuse was associated
with truncated daily (episodes of 1–3 days) and hourly
(episodes of ,24 hours) cycling, particularly in patients in
whom substance abuse began prior to the onset of the
bipolar disorder. Recently, Goldberg et al (1999) also
found that a history of substance abuse was associated
with lower rates of remission during hospitalization, as
well as poor lithium treatment response.
A significant confound in many of these studies is the
inclusion of patients with different histories of past affective episodes. Since the number of previous affective
episodes becomes a strong predictor of future affective

course, this may make it difficult to identify other course
predictors (Strakowski et al 1998a). Studies of firstepisode patients address this problem by removing the
confound of illness chronicity. Strakowski et al (1993)
studied the effects of psychiatric comorbidity in a sample
of 60 bipolar patients with a first-episode of psychotic

BIOL PSYCHIATRY
2000;48:477– 485

479

mania. They observed that psychiatric comorbidity was
associated with poorer rates of recovery during hospitalization; this finding was limited by including a number of
substance use and psychiatric disorders in the definition of
psychiatric comorbidity. Subsequently, in a 2-year follow-up of these patients (plus additional subjects recruited
later) with first-episode affective psychosis (n 5 219),
substance abuse was not significantly associated with
outcome after controlling for other variables (e.g., index
depression score, brief initial hospitalization, age of onset
after age 30 years) (Tohen et al 2000). These results were
confounded, however, by the combination of patients with
first-episode (unipolar) psychotic depression and those
with bipolar disorder.
Strakowski et al (1998a) extended prior work by prospectively examining the effects of interval substance
abuse, i.e., substance abuse during the follow-up period,
rather than a lifetime history of substance abuse in 109
patients with first-episode affective psychoses including
83 patients with bipolar disorder. Most prior studies
combined patients with any lifetime history of substance
abuse into a single group, thereby including patients
actively abusing drugs and alcohol with those that had not
used substances for many years. In this study, interval
substance abuse was associated with impaired symptomatic recovery (i.e., resolution of essentially all affective
symptoms for at least 8 weeks) in a Cox regression model
controlling for other relevant clinical variables. Although
interval substance use did not also predict impaired syndromic recovery (i.e., the resolution of enough symptoms
that criteria for a DSM-III-R affective syndrome were no
longer met), it was associated with treatment compliance,
which was the strongest predictor of syndromic recovery.
Therefore, the authors concluded that interval substance
abuse had both direct and indirect effects on the course of
new onset bipolar and psychotic depressive disorders.
Namely, substance abuse was associated with a failure of
affective symptoms to resolve, which appeared to be a
direct effect, and also contributed to poor treatment
compliance, leading indirectly to impaired recovery. Similar findings were observed in a multiple-episode bipolar
sample from the same research group (Keck et al 1998).
Other investigators examined interactions between the
courses of substance use and bipolar disorders in more
detail, looking at variables other than simple recovery and
outcome. Young et al (1981) followed 15 bipolar patients
with alcohol abuse in a lithium treatment trial for 1 year.
They found that symptoms of alcoholism tended to appear
before affective symptoms in those patients with relapse of
their bipolar illness. The NIMH Collaborative Study on
the Psychobiology of Depression (CDS) recruited 231
patients with bipolar or schizoaffective mania and followed them every 6 months for 5 years then annually after

480

BIOL PSYCHIATRY
2000;48:477– 485

that using the Longitudinal Follow-up Evaluation (LIFE;
Keller et al 1987). Seventy (30%) of these patients had a
lifetime history of alcoholism defined by Research Diagnostic Criteria (RDC; Endicott and Spitzer 1978), including 34 patients in whom the alcoholism developed first
(“primary alcoholism”), 30 in whom the bipolar disorder
developed first (“primary bipolar disorder”), and six in
whom these temporal relationships could not be determined. In this study, Winokur et al (1994, 1995) found that
patients with primary bipolar disorder had significantly
more affective episodes after the initial recovery, and the
median time to their first affective relapse was shorter as
compared to the bipolar patients with primary alcoholism.
The patients with primary bipolar disorder were also less
likely to recover during follow-up. Notably, the types of
symptoms of alcoholism did not differ between these two
groups of bipolar patients. Additionally, in contrast to a
comparison group of patients with primary alcoholism and
secondary depression, the bipolar patients with alcoholism
(primary, secondary, or undefined) were significantly less
likely to exhibit persistent symptoms of alcohol abuse
during the first five years of follow-up. Specifically, by the
5-year follow-up only 5% of the bipolar patients, compared to 25% of unipolar patients, continued to exhibit
RDC alcoholism.
Recently, we extended these findings by prospectively
examining associations between substance use and bipolar
disorders during the 12 months after a first psychiatric
hospitalization for psychotic mania (Strakowski et al
1998b). In these 77 patients, 26 (34%) patients had a
history of alcohol use disorders and 28 (36%) had a drug
use disorder. During the follow-up interval, 54% of the
patients with an alcohol use disorder and 39% of patients
with a drug use disorder experienced affective episodes in
the absence of ongoing alcohol or drug abuse. Therefore,
patients whose bipolar illness co-occurs with substance
abuse do not necessarily require ongoing substance abuse
to initiate new affective episodes; however, whenever
present, alcohol abuse was associated with a current
affective syndrome. In contrast, 21% of these patients
continued to abuse drugs (cannabis in all cases) and yet
maintained remission of their bipolar symptoms. This
observation suggests that the interactions between different substances of abuse and bipolar disorder may differ.
Specifically, alcohol appeared to be strongly associated
with maintaining or developing affective symptoms,
whereas this may be less true with cannabis. Clearly, these
results need to be replicated in larger patient samples over
longer periods of time. Finally, only one patient with no
prior history of an alcohol use disorder developed a new
alcohol abuse syndrome during the follow-up interval and
no patients developed a new drug abuse syndrome. Therefore, at least in this relatively short follow-up period,

S.M. Strakowski et al

developing new cases of substance use disorders after the
first manic episode were rare. Coupled with the findings
from Winokur et al (1995), these results suggest that once
patients are in treatment (or at least in follow-up studies)
their risk of developing new substance abuse syndromes or
maintaining substance use disorders appears to decrease
over time.

Bipolar Disorder May Impact the Course of
Substance Abuse
In addition to the effects of substance abuse on the course
of bipolar disorder, there are a few studies that have
examined how bipolar disorder may influence substance
abuse syndromes (reviewed in Strakowski and DelBello
2000). For example, in a sample of 59 bipolar patients,
Mayfield and Coleman (1968) found that 32% increased
their alcohol use when manic, whereas only 10% increased
their drinking when depressed. More commonly, patients
decreased their drinking when depressed (17%), although,
in most cases, alcohol consumption did not change with
affective state. A number of other authors have reported
similar findings (Bernadt and Murray 1986; Dunner et al
1979; Hensel et al 1979; Reich et al 1974; Winokur et al
1969). Taken together, these studies suggest that most
patients (approximately three fourths) do not change their
alcohol consumption during a manic episode, although
some patients increase their alcohol use over baseline and
only rarely do patients decrease their drinking in this
affective state. In contrast, during depression, patients are
as likely to decrease as to increase their drinking (approximately 15% of each), although, again, most do not change
their alcohol use. There are no quantitative studies of
changes in drug use (other than alcohol) during affective
episodes in bipolar patients. Notably, none of these studies
provided comparison groups (e.g., patients with primary
substance use disorders) to control for the changes over
similar time intervals that occur in alcohol or drug use in
persons without bipolar disorder.
Despite these previous studies, many aspects of the
relationships between the courses of substance use and
bipolar disorders remain unknown. For example, it has not
been clarified whether alcohol or drug use disorders are
specifically associated with more time spent in manic or
depressive episodes during the course of bipolar illness.
Additionally, as noted, methods for defining subgroups of
bipolar patients with different types of associations between the courses of substance use and bipolar disorders
have not been developed. With these considerations in
mind, we initiated the University of Cincinnati First
Episode Mania study in June 1996. The aims of this
ongoing study are to prospectively examine the relative
courses of bipolar and substance use disorders in patients

Substance Abuse in Bipolar Disorder

BIOL PSYCHIATRY
2000;48:477– 485

481

recruited at the time of their first psychiatric hospitalization for mania. By incorporating the methodology of the
NIMH CDS, we generate week-by-week course of affective and substance abuse symptoms. We examined our
data in order to address two questions relevant to this
review:

SCID-P in conjunction with completing the Addictions Severity
Index (ASI; McClellan et al 1992). The investigators demonstrate good interrater reliability for symptom measures (intraclass
correlation coefficient . .70).

1. Is the duration of time with alcohol or cannabis use
disorders associated with the time spent in manic or
depressive episodes during the course of bipolar
disorder?
2. Can correlations in the courses of substance use and
bipolar disorders define subgroups of patients with
differing relationships between these co-occurring
conditions?

Demographic information was obtained from direct patient
interviews and review of medical records. This included gender,
race, years of education, current age, and employment status
prior to the onset of the current affective episode.

Methods and Materials
The methods of this study have been described in detail elsewhere (Strakowski et al, in press), but they are presented briefly
here.

Subjects
Recruitment for this study began June 1, 1996. Inclusion criteria
include: 1) meets DSM-IV criteria for bipolar disorder, manic or
mixed; 2) age 16 – 45 years; 3) no prior psychiatric hospitalizations; 4) less than 1 month prior psychotropic medication; 5) able
to communicate in English; 6) lives within 50 miles of Cincinnati; and 7) provision of written informed consent after study
procedures have been fully explained and understood. Patients
are excluded if psychiatric symptoms: 1) are entirely due to acute
medical illness as determined by examination; 2) result from
acute intoxication or withdrawal from drugs or alcohol as
determined by symptom resolution within the expected period of
acute intoxication or withdrawal; or 3) if patients had diagnosed
mental retardation (i.e., IQ , 70). This article evaluates the first
50 consecutively enrolled subjects who had at least 16 weeks of
follow-up. We plan to recruit 150 –200 total subjects for this
study. Patients both with and without lifetime histories of
substance use disorders are included. A lifetime alcohol or drug
use disorder was present in 66% (n 5 33) of this sample, and 23
(46%) met criteria for an alcohol or drug use disorder during the
month prior to the index hospitalization.

Index Clinical Assessment
The diagnosis of DSM-IV bipolar disorder (manic or mixed) was
established by psychiatrists or psychologists using the Structured
Clinical Interview for DSM-IV, Patient versions (SCID-P; k .
.90) (First et al 1995). Psychiatric symptoms were assessed using
the Young Mania rating scale (YMRS; Young et al 1978), the
17-item Hamilton Depression Rating Scale (HDRS; Hamilton
1960), and the Scale for the Assessment of Positive Symptoms
(SAPS; Andreasen 1984). Substance use disorder assessments
were made by trained research assistants supervised by a psychiatrist using the Substance Use Disorders module of the

Demographic Variables

Follow-Up Assessments
After hospital discharge, patients are evaluated at 1 month and 4
months, then every 4 months for as long as this study remains
funded. For this analysis, the maximal follow-up period analyzed
was 24 months (104 weeks). For the subjects used in this
analysis, the mean length of follow-up was 66 (SD 5 32) weeks
(range 16 –104 weeks).
At each follow-up visit, investigators review, week-by-week,
the prior interval. Particular attention is paid to times of symptom
change as with the CDS (Keller et al 1987) and our previous
work (Keck et al 1998; Strakowski et al 1998a). This follow-up
review includes each item of the affective and substance abuse
symptom ratings scales (YMRS, HDRS, SAPS, and ASI) in
conjunction with the SCID-P for that interval. From these
ratings, week-by-week six-point ratings are made of the severity
of affective and substance abuse syndromes (Table 1). Then the
percent of weeks with full affective or substance abuse syndromes (scores of 5 or 6) or significant symptoms (scores of 3 or
4) was calculated. Random urine toxicology and alcohol breathalyzer screens are obtained to verify self-reports of substance
abuse.

Treatment Assessments
Although this is a naturalistic study, the treatments patient
receive during follow-up are recorded. Using these data, treatment compliance was assessed as 1) full compliance, in which
pharmacologic treatment was taken more than 75% as prescribed; 2) total noncompliance, in which pharmacologic treatment was taken less than 25% of the time as prescribed; and 3)
partial noncompliance, in which pharmacologic treatment was
taken between these two extremes (Keck et al 1998; Strakowski
et al 1998a). This rating was obtained by reviewing interval
medication use with each patient, family members and clinicians,
when indicated (i.e., if a patient’s reliability is suspect). From
this review, the percent of follow-up in which patients exhibited
each category of compliance was determined for each prescribed
psychotropic medication and an average score across medications was then used for analysis.

Statistical Analysis
To answer question 1, a regression model was evaluated examining associations between the fraction of time during follow-up
in which patients experienced full affective syndromes or signif-

482

S.M. Strakowski et al

BIOL PSYCHIATRY
2000;48:477– 485

Table 1. Psychiatric Status Ratings—Determination of
Symptomatic Recovery
Code

Term

Definition

6

Full syndrome, severe

5

Full syndrome

4

Marked symptoms

3

Partial remission

Meets DSM-IV criteria for severe
syndrome
Meets DSM-IV criteria for mild
to moderate syndrome
Does not meet DSM-IV criteria
for syndrome, but one or more
DSM-IV syndrome criterion
items are scored greater than
“mild”
No DSM-IV syndrome criterion
item scored greater than “mild”
but:
For affective disorder: YMRS
total score . 5, HAM-D total
score . 7, or any SAPS global
item scores . 2
For substance abuse disorder:
ASI substance use score . 2
Has no DSM-IV syndrome
criterion item scored greater
than “mild” and:
For affective disorder: YMRS
total score # 5, HAM-D total
score # 7, and SAPS global
scores all # 2
For substance abuse disorder:
ASI substance use score , 2
No affective, psychotic or
substance abuse symptoms

2

1

Residual symptoms

Usual self

YMRS, Young Mania Rating Scale; HAM-D, Hamilton Depression Rating
Scale; SAPS, Scale for the Assessment of Positive Symptons; ASI, Addictions
Severity Index.

icant affective symptoms and the fraction of time with significant
cannabis or alcohol abuse symptoms or syndromes (only cannabis and alcohol abuse occurred at high enough rates in this
sample to permit analysis, as no other drug was abused by the
patients during the follow-up interval). Both alcohol and cannabis abuse symptoms/syndromes were included in a single model
to control for the interaction. Potential confounds were identified
using simple statistics (t tests and x2) and variables that demonstrated associations with cannabis or alcohol abuse symptoms at
a liberal p , .2 level were entered into the regression model.
Variables examined as confounds included age, gender, race,
education, employment level, affective state (mixed vs. manic),
age of bipolar disorder onset, duration of the index episode, and
treatment noncompliance.
To answer question 2, cross-correlations using Kendall t-b
were used to compare the weekly time series of severity of
affective symptom/syndrome ratings with cannabis and alcohol
symptom/syndrome ratings for each patient. The time series were
right- or left-shifted (from 1 to 8 weeks) relative to each other to
identify whether the maximal correlation between time series
demonstrated a temporal delay or advance. Specifically, if the
maximum correlation occurred with the cannabis or alcohol
time-series shifted to the left, then that suggested that substance

Table 2. Demographic and Clinical Characteristic of 50
Patients with Bipolar Disorder Following a First Psychiatric
Hospitalization According to Whether They Exhibited Alcohol
or Cannabis Abuse during the Follow-Up Interval

Characteristic
Age (Years)
Race [n (% nonwhite)]
Gender [n (% women)]
Education (years)
Unemployed [n (%)]
Index mixed state [n (%)]
Age at bipolar onset (years)
Duration of index episode
(weeks)
Index lifetime substance use
disorder
Any substance
Alcohol [n (%)]
Cannabis [n (%)]
Cocaine [n (%)]
Hallucinogens [n (%)]
Sedatives [n (%)]
Other stimulants [n (%)]
Opiates [n (%)]

Interval
abuser
(n 5 21)

Interval
nonabuser
(n 5 29)

Total
(n 5 50)

25 (7)
10 (48)
5 (24)
12 (2)
6 (29)
6 (29)
22 (7)
7 (7)

25 (9)
12 (41)
13 (45)
12 (3)
11 (38)
14 (48)
22 (8)
15 (23)

25 (8)
22 (44)
18 (36)
12 (3)
17 (34)
20 (40)
22 (8)
12 (19)

20 (95)a,b
15 (71)c
16 (76)d
2 (10)
3 (14)
0 (0)
1 (5)
0 (0)

13 (45)
9 (31)
10 (34)
3 (10)
2 (7)
3 (10)
1 (3)
1 (3)

33 (66)
24 (48)
26 (52)
5 (10)
5 (10)
3 (6)
2 (4)
1 (2)

a
One subject had a history of cannabis use that did not meet abuse or
dependence criteria before the index hospitalization that worsened during followup.
b
Significant difference from interval nonabusers: x2(1) 5 13.8, p , .001.
c
Significant difference from interval nonabusers: x2(1) 5 8.0, p , .005.
d
Significant difference from interval nonabusers: x2(1) 5 8.5, p , .004.

abuse may have precipitated the affective episode. In contrast, if
the maximum correlation occurred with the substance abuse time
series shifted to the right, then this relationship suggests that
substance abuse occurred as a consequence of the affective
relapse. A t-b 5 .20 was defined as having a clinically
meaningful association. Finally, paired t tests were used to
determine whether the overall group cross-correlations significantly deviated from zero.

Results
Demographic and clinical characteristics of this sample
are listed in Table 2. During follow-up, the patients spent
a mean of 33% (SD 5 36%) of the time totally noncompliant with treatment. They experienced full affective
syndromes 25% (SD 5 25%) of follow-up, including 10%
(SD 5 18%) of the time with mania, 12% (SD 5 22%)
with depression, and 3% (SD 5 9%) in a mixed state. An
additional 27% (SD 5 28%) of follow-up was spent with
significant affective symptoms that did not meet full
syndrome criteria. The sample exhibited symptoms or a
full syndrome of an alcohol use disorder 11% (SD 5 32%)
of follow-up. Similarly, they exhibited cannabis abuse
symptoms or syndromes for 13% (SD 5 31%) of the time.
Index affective state (i.e., mixed or manic), gender, and

Substance Abuse in Bipolar Disorder

BIOL PSYCHIATRY
2000;48:477– 485

Table 3. Associations between Changes in Affective
Symptoms and Changes in Alcohol and Cannabis Abuse
Following a First Psychiatric Hospitalization in 50 Patients
with Bipolar Disorder
Substance
Alcohol [n (%)]
Mean t-b (SD)
Cannabis [n (%)]
Mean t-b (SD)

Left
shifted

Right
shifted

7 (44)
1 (6)
.39 (.21) .88 (0)
5 (26)
4 (21)
.52 (.13) .53 (.17)

Negative

None

6 (38)
2.35 (.08)
7 (37)
2.51 (.23)

2 (12)
.03 (.23)
3 (16)
2.08 (.14)

Listed are the number of subjects exhibiting maximal correlations with
substance abuse left-shifted relative to affective symptoms (increased substance
abuse leads to increased affective symptoms), with substance abuse right-shifted
(increased substance abuse follows increased affective symptoms), with negative
associations (changes in substance abuse occur inversely to changes in affective
symptoms) and with no association (defined as t-b , 0.20).

483

cannabis abuse. For both alcohol and cannabis abuse,
negative associations with affective symptoms were also
common (Table 3); however, at none of the time shifts did
the sample, as a whole, exhibit a statistically significant
overall pattern of association between alcohol use or
cannabis use and affective symptoms/syndromes. In contrast, alcohol and cannabis use exhibited significant associations with each other, with the strongest relationship
present when alcohol was left-shifted 4 weeks (t 5 .31,
SD 5 .32, t 5 3.2, p 5 .009), suggesting alcohol use
increases prior to increases in cannabis use in these
patients.

Discussion
fraction of time with noncompliance were associated with
either interval cannabis or alcohol ratings (at p , .2) and
so were entered as potential confounds in the regression
model. In this model, the fraction of time with alcohol
abuse symptoms/syndromes was significantly associated
with the fraction of time with any affective syndrome
(adjusted partial R 5 .37, p 5 .01) after controlling for the
fraction of time with cannabis abuse symptoms/syndromes
and the confounds. (All subsequent R values reported in
this section are likewise partial scores adjusted for confounds and the other substance use disorder.) Most of this
association was due to correlations between the fraction of
time with alcohol symptoms/syndromes and the fraction of
time spent in depression (R 5 .33, p 5 .025) rather than
mania (R 5 .17, p . .2) or mixed states (R 5 .11, p . .4).
In contrast, the fraction of time with cannabis abuse
symptoms/syndromes was not significantly associated
with the fraction of time in depression (R 5 .21, p . .1)
nor in mixed states (R , .02, p . .9), but was significantly
associated with the fraction of time with mania (R 5 .42,
p 5 .004). The fraction of time with alcohol abuse
symptoms/syndromes was significantly correlated with the
time with cannabis abuse symptoms/syndromes (R 5 .49,
p 5 .0005).
Sixteen patients exhibited changes in both alcohol use
and affective symptom/syndrome ratings scores during
follow-up to permit evaluation of associations between
changes in these two disorders using the correlative
analysis of the respective time series. As illustrated in
Table 3, the most common association was with the
alcohol use series left-shifted 1– 8 weeks (n 5 7, 44%).
Only a single patient exhibited the opposite pattern with
affective symptoms preceding increased alcohol use by 1
week. Similarly, 19 patients exhibited changes in both
cannabis use and affective ratings to evaluate these temporal relationships. In contrast to alcohol use, similar
numbers of subjects exhibited both left- and right-shifted
maximal associations between affective symptoms and

From this review, it appears that alcohol use disorders may
affect the age of bipolar disorder onset. In particular,
bipolar patients with antecedent alcohol abuse appear to
have a later age of onset of bipolar illness, which suggests
that affective episodes in those patients are not precipitated until after several years of alcohol abuse. This
subgroup of patients may have a lower familial risk of
developing bipolar disorder than bipolar patients without
alcohol abuse (DelBello et al 1999), and may therefore
“require” the alcohol abuse to initiate the illness (Winokur
et al 1995); however, many patients with alcohol abuse,
even when it is antecedent to the bipolar disorder, will
continue to experience affective episodes even after
achieving sobriety (Strakowski et al 1998b), suggesting
ongoing alcohol abuse is not required to maintain the
affective symptoms. Nevertheless, once the courses of
bipolar and alcohol use disorders are established, the
longer term prognosis suggests a resolution of the alcohol
problems in most patients. Once sobriety is achieved,
patients with a past history of both a substance use and
bipolar disorder may subsequently experience a better
course of illness than patients without such a history of
co-occurrence (Winokur et al 1995). Nonetheless, alcohol
and drug abuse are associated with impaired treatment
response, and symptomatic and functional recovery in
bipolar disorder. These effects may be direct, because
alcohol abuse may specifically precipitate affective symptoms (Strakowski et al 1998a, 1998b) or indirect by
interfering with treatment compliance (Strakowski et al
1998a). The new results we present here suggest that
alcohol abuse is primarily associated with depression,
whereas cannabis abuse may be more commonly associated with mania. Additionally, these associations do not
exhibit a single, common pattern in all patients, but
instead, there appear to be subgroups of patients who
exhibit different temporal relationships among affective
symptomatology and cannabis and alcohol abuse. Moreover, different substances of abuse may be associated with

484

BIOL PSYCHIATRY
2000;48:477– 485

different patterns. The common finding that affective
symptoms and alcohol and cannabis abuse are inversely
associated suggests that either substance use masks or
decreases affective symptoms or, conversely, that affective symptoms may result from substance withdrawal.
These suggestions required additional study as, clearly,
these new data should be viewed as preliminary, since the
number of subjects is still relatively small.
Most previous studies examining the common cooccurrence between substance use and bipolar disorders
have been limited to simply reporting elevated rates of
substance abuse or have been post-hoc analyses examining
outcome associations. Few studies have been designed
specifically to detail the relationships among these conditions. Although these relationships are complex, the results we present here suggest they can be examined
systematically and, hopefully, clarified. Understanding the
relationships between substance use and bipolar disorders
may lead to new treatment approaches for patients with
both conditions (Strakowski and DelBello 2000). Moreover, clarifying how different substances of abuse differentially impact the course of bipolar disorder might help
define the neurophysiology of bipolar disorder.

Supported by an NIMH award MH58170 (SMS).
Aspects of this work were presented at the conference “Bipolar
Disorder: From Preclinical to Clinical, Facing the New Millennium,”
January 19 –21, 2000, Scottsdale, Arizona. The conference was sponsored by the Society of Biological Psychiatry through an unrestricted
educational grant provided by Eli Lilly and Company.

References
American Psychiatric Association (1994): Diagnostic and Statistical Manual of Mental Disorders, 4th ed. Washington, DC:
APA Press.
Andreasen NC (1984): The Scale for The Assessment of Positive
Symptoms (SAPS). Iowa City: University of Iowa.
Berkson J (1946): Limitations of four-fold tables to hospital data.
Biometrics Bull 35:47–53.
Bernadt MW, Murray, RM (1986). Psychiatric disorder, drinking
and alcoholism: What are the links? Br J Psychiatry 148:393–
400.
DelBello MP, Strakowski SM (in press): The co-occurrence of
affective and substance use disorders. In: Westermeyer J,
Weiss R, editors. Integrating Treatment for Mood and Substance Disorder. Baltimore: Johns Hopkins University Press.
DelBello MP, Strakowski SM, Sax KW, McElroy SL, Keck PE
Jr, West SA, Kmetz GF (1999): Effects of familial rates of
affective illness and substance abuse on rates of substance
abuse in patients with first-episode mania. J Affect Disord
56:55– 60.
Dunner DL, Hensel BM, Fieve RR (1979): Bipolar illness:
Factors in drinking behavior. Am J Psychiatry 136:583–585.
Endicott J, Spitzer RL (1978): A diagnostic interview: The

S.M. Strakowski et al

Schedule for Affective Disorders and Schizophrenia. Arch
Gen Psychiatry 35:837– 844.
Feinman JA, Dunner DL (1996): The effect of alcohol and
substance abuse on the course of bipolar affective disorder. J
Affect Disord 37:43– 49.
First MB, Spitzer RL, Gibbon M, Williams JBW (1995):
Structured Clinical Interview for DSM-IV Axis I Disorders—
Patient Edition (SCID-I/P, version 2.0). New York: New
York State Psychiatric Institute, Biometrics Research Department.
Fogarty F, Russell JM, Newman SC, Bland RC (1994): Epidemiology of psychiatric disorders in Edmonton. Mania. Acta
Psychiatr Scand Suppl 376:16 –23.
Goldberg JF, Garno JL, Leon AC, Kocsis JH, Portera L (1999):
A history of substance abuse complicates remission from
acute mania in bipolar disorder. J Clin Psychiatry 60:733–
740.
Hamilton M (1960): A rating scale for depression. J Neurol
Neurosurg Psychiatry 23:56 – 61.
Hensel B, Dunner DL, Fieve RR (1979): The relationship of
family history of alcoholism to primary affective disorder. J
Affect Disord 1:105–113.
Himmelhoch JM, Mulla D, Neil JF, Detre TP, Kupfer DJ (1976):
Incidence and significance of mixed affective states in a
bipolar population. Arch Gen Psychiatry 33:1062–1066.
Keck PE Jr, McElroy SL, Strakowski SM, West SA, Sax KW,
Hawkins JM, et al (1998): Twelve-month outcome of bipolar
patients following hospitalization for a manic or mixed
episode. Am J Psychiatry 155:646 – 652.
Keller MB, Lavori PW, Griedman B, Nielsen E, Endicott J,
McDonald-Scott P, Andreasen NC (1987): The Longitudinal
Interval Follow-up Evaluation: A comprehensive method for
assessing outcome in prospective longitudinal studies. Arch
Gen Psychiatry 44:540 –548.
Kessler RC, Crum RM, Warner LA, Nelson CB, Schulenberg J,
Anthony JC (1997): Lifetime co-occurrence of DSM-III-R
alcohol abuse and dependence with other psychiatric disorders in the National Comorbidity Survey. Arch Gen Psychiatry 54:313–321.
Mayfield DG, Coleman LL (1968): Alcohol use and affective
disorder. Dis Nerv Syst 29:467– 474.
McClellan AT, Kushner H, Metzger D, Peters R, Smith I,
Grissom G, et al (1992): The Fifth Edition of the Addiction
Severity Index. J Subst Abuse Treat 9:199 –213.
Morrison JR (1974): Bipolar affective disorder and alcoholism.
Am J Psychiatry 131:1130 –1133.
Mueser KT, Drake RE, Wallach MA (1998): Dual diagnosis: A
review of etiological theories. Addict Behav 23:717–734.
Post RM (1992): Transduction of psychosocial stress into the
neurobiology of recurrent affective disorder. Am J Psychiatry
149:999 –1010.
Regier DA, Farmer ME, Rae DS, Locke BZ, Keith SJ, Judd LL,
Goodwin FK (1990): Comorbidity of mental disorders with
alcohol and other drug abuse: Results from the Epidemiologic
Catchment Area (ECA) Study. JAMA 264:2511–2518.
Reich LH, Davies RK, Himmelhoch JM (1974): Excessive
alcohol use in manic-depressive illness. Am J Psychiatry
131:83– 86.
Sonne SC, Brady KT, Morton WA (1994): Substance abuse and
bipolar affective disorder. J Nerv Ment Dis 182:349 –352.

Substance Abuse in Bipolar Disorder

Strakowski SM, DelBello MP (2000): The co-occurrence of
substance use and bipolar disorders. Clin Psychol Rev 20:
191–206.
Strakowski SM, Keck PE Jr, McElroy SL, West SA, Sax KW,
Hawkins JM, et al (1998a): Twelve-month outcome after a
first hospitalization for affective psychosis. Arch Gen Psychiatry 55:49 –55.
Strakowski SM, McElroy SL, Keck PE Jr, West SA (1994): The
co-occurrence of mania with medical and other psychiatric
disorders. Int J Psychiatry Med 24:305–328.
Strakowski SM, McElroy SL, Keck PE Jr, West SA (1996a): The
effects of antecedent substance abuse on the development of
first-episode mania. J Psychiatr Res 30:59 – 68.
Strakowski SM, Sax KW, McElroy SL, Keck PE Jr, Hawkins
JM, West SA (1998b): Psychiatric and substance abuse
syndrome co-occurrence in bipolar disorder following a first
psychiatric hospitalization. J Clin Psychiatry 59:465– 471.
Strakowski SM, Sax KW, Setters MJ, Keck PE Jr (1996b):
Enhanced response to repeated d-amphetamine challenge:
Evidence for behavioral sensitization in humans. Biol Psychiatry 40:872– 880.
Strakowski SM, Tohen M, Stoll AL, Faedda GL, Mayer PV,
Kolbrener ML, Goodwin DC (1993): Comorbidity in psychosis at first hospitalization. Am J Psychiatry 150:752–757.
Strakowski SM, Williams JR, Sax KW, Fleck DE, DelBello MP,
Bourne ML (in press): Is impaired outcome following a first
manic episode due to mood-incongruent psychosis? J Affect
Disord.
Tohen M, Hennen J, Zarate CM Jr, Baldessarini RJ, Strakowski
SM, Stoll AL, et al (2000): The McLean First Episode
Project: Two-year syndromal and functional recovery in 219
cases of major affective disorders with psychotic features.
Am J Psychiatry 157:220 –228.
Tohen M, Waternaux CM, Tsuang MT (1990): Outcome in

BIOL PSYCHIATRY
2000;48:477– 485

485

mania: A 4-year prospective follow-up of 75 patients using
survival analysis. Arch Gen Psychiatry 47:1106 –1111.
Tsai SY, Chen CC, Hu WH, Lee JC, Chao WS, Yeh EK (1996):
Comorbidity of substance abuse in patients with bipolar
disorder: A 15-year follow-up study. Taiwan J Psychiatry
10:357–364.
Tsai SY, Chen CC, Kuo CJ, Lee JC, Lee HC, Strakowski SM (in
press): Fifteen-year outcome of treated bipolar disorder. J
Affect Disord.
Verdoux H, Mury M, Besancon G, Bourgeois M (1996): Comparative study of substance dependence comorbidity in bipolar, schizophrenic and schizoaffective disorders. Encephale
22:95–101.
Winokur G, Clayton PJ, Reich T (1969): Manic Depressive
Illness. St. Louis: Mosby.
Winokur G, Coryell W, Akiskal HS, Endicott J, Keller M,
Mueller T (1994): Manic-depressive (bipolar) disorder: The
course in light of a prospective ten-year follow-up of 131
patients. Acta Psychiatr Scand 89:102–110.
Winokur G, Coryell W, Akiskal HS, Maser JD, Keller MB,
Endicott J, Mueller T (1995): Alcoholism in manic-depressive (bipolar) illness: Familial illness, course of illness, and
the primary-secondary distinction. Am J Psychiatry 152:365–
372.
Winokur G, Coryell W, Endicott J, Keller M, Akiskal H,
Solomon D (1996): Familial alcoholism in manic-depressive
(bipolar) disease. Am J Med Genet 67:197–201.
Young LD, Patel M, Keeler MH (1981): The effect of lithium
carbonate on alcoholism in 20 male patients with concurrent
major affective disorder. Curr Alcohol 8:175–181.
Young RC, Biggs JT, Ziegler VE, Meyer DA (1978): A rating
scale for mania: Reliability validity and sensitivity. Br J
Psychiatry 133:429 – 435.