Original Investigation | Psychiatry

Longitudinal Interplay Between Alcohol Use, Mood, and Functioning

in Bipolar Spectrum Disorders
Sarah H. Sperry, PhD; Audrey R. Stromberg, BA; Victoria A. Murphy, BS; Carly A. Lasagna, MS; Melvin G. McInnis, MD; Margo W. Menkes, MS;
Anastasia K. Yocum, PhD; Ivy F. Tso, PhD

Abstract Key Points

Question How is alcohol use associated
IMPORTANCE Alcohol use disorder (AUD) is present in nearly half of individuals with bipolar
with mood and life functioning over the
disorder (BD) and is associated with markedly worsening outcomes. Yet, the concurrent treatment
longitudinal course of bipolar
of BD and AUD remains neglected in both research and clinical care; characterizing their dynamic
disorder (BD)?
interplay is crucial in improving outcomes.
Findings In this cohort study of 584
OBJECTIVE To characterize the longitudinal alcohol use patterns in BD and examine the temporal adults with BD, increased alcohol use
associations among alcohol use, mood, anxiety, and functioning over time. was associated with increased
depression and manic or hypomanic
DESIGN, SETTING, AND PARTICIPANTS This cohort study selected participants and analyzed data symptoms as well as lower workplace
from the Prechter Longitudinal Study of Bipolar Disorder (PLS-BD), an ongoing cohort study that functioning over the following 6
recruits through psychiatric clinics, mental health centers, and community outreach events across months, but not vice versa. Associations
Michigan and collects repeated phenotypic data. Participants selected for the present study were were more pronounced in BD type II
those with a diagnosis of BD type I (BDI) or type II (BDII) who had been in the study for at least 5 than BD type I.
years. Data used were extracted from February 2006 to April 2022, and follow-up ranged from 5 to
Meaning Findings of this study suggest
16 years.
there is an association of alcohol use
with mood and work functioning,
MAIN OUTCOMES AND MEASURES Alcohol use was measured using the Alcohol Use Disorders
highlighting the importance of
Identification Test. Depression, mania or hypomania, anxiety, and functioning were measured using
dimensional and longitudinal
the 9-Item Patient Health Questionnaire, the Altman Self-Rating Mania Scale, the 7-item Generalized
assessment and management of alcohol
Anxiety Disorder assessment scale, and the Life Functioning Questionnaire, respectively.
use, which should be integrated into
research and standard treatment of BD.
RESULTS A total of 584 individuals (386 females (66.1%); mean [SD] age, 40 [13.6] years) were
included. These participants had a BDI (445 [76.2%]) or BDII (139 [23.8%]) diagnosis, with or without
a lifetime diagnosis of AUD, and a median (IQR) follow-up of 9 (0-16) years. More problematic alcohol + Supplemental content
use was associated with worse depressive (β = 0.04; 95% credibility interval [CrI], 0.01-0.07) and Author affiliations and article information are
manic or hypomanic symptoms (β = 0.04; 95% CrI, 0.01-0.07) as well as lower workplace listed at the end of this article.

functioning (β = 0.03; 95% CrI, 0.00-0.06) over the next 6 months, but increased depressive and
manic or hypomanic symptoms were not associated with greater subsequent alcohol use. These
latter 2 associations were more pronounced in BDII than BDI (mania or hypomania: β = 0.16 [95% CrI,
0.02-0.30]; workplace functioning: β = 0.26 [95% CrI, 0.06-0.45]). Alcohol use was not associated
with anxiety over time.

CONCLUSIONS AND RELEVANCE This study found that alcohol use, regardless of diagnostic status,
was associated with mood instability and poorer work functioning in BD, but increased mood
symptoms were not associated with subsequent alcohol use. Given its prevalence and repercussions,
dimensional and longitudinal assessment and management of alcohol use are necessary and should
be integrated into research and standard treatment of BD.

JAMA Network Open. 2024;7(6):e2415295. doi:10.1001/jamanetworkopen.2024.15295

Open Access. This is an open access article distributed under the terms of the CC-BY License.

JAMA Network Open. 2024;7(6):e2415295. doi:10.1001/jamanetworkopen.2024.15295 (Reprinted) June 7, 2024 1/14

Downloaded from jamanetwork.com by guest on 06/18/2024

 JAMA Network Open | Psychiatry Interplay Between Alcohol Use, Mood, and Functioning in Bipolar Spectrum Disorders

Introduction
Up to 45% of individuals with bipolar disorder (BD) have co-occurring alcohol use disorder (AUD).1-3
Although BD and AUD are associated with poor outcomes,4,5 their co-occurrence may further
exacerbate risk.6,7 Yet, current treatments for BD rarely include or consider AUD management, and
problematic alcohol use is often excluded in BD trials.6 Consequently, the research informing clinical
knowledge lacks representation from nearly half of the study population, and the extent of the
interplay between alcohol use and BD symptoms is largely unexplored.
Prior cross-sectional work highlights that co-occurrence of AUD and BD is associated with
prolonged alcohol withdrawal, more expensive treatment, decreased functioning, increased
suicidality, and increased morbidity.1,5,8 These findings are generally corroborated by longitudinal
studies. In a 4-year longitudinal study of BD type I (BDI), a history of AUD was associated with poor
recovery after mania.8 In a 5-year follow-up, individuals with BDI and AUD experienced more suicidal
behavior and poorer social functioning compared with those without AUD.9 Furthermore, those with
BD type II (BDII) and AUD were more likely to have a manic episode and transition to a BDI diagnosis
5 years later, highlighting that co-occurring AUD is associated with worse outcomes of BD regardless
of subtype.
Reviews also highlight associations between AUD and more complex (eg, mixed or dysphoric
mania2,10 and rapid cycling2,10-12) and severe mood symptoms,1,6,7,13-15 the onset of which can be
precipitated by alcohol use.12,16,17 Depressive symptoms seem to both precede and occur after
alcohol use in BD throughout follow-up periods ranging from 8 months13 to 10 years.18 The
association of anxiety, which affects up to 45% of individuals with BD,19 with alcohol use in BD has
been less studied. Given the bidirectional risk between AUD and anxiety,20 this gap in the BD
literature poses considerable clinical implications.
Despite the literature, it remains unclear how alcohol use fluctuates over time in BD and how
longitudinal dynamics interact with proximal changes in depressive, manic or hypomanic, and
anxiety symptoms. Examining these dynamics can inform the mechanisms of how alcohol use plays
a role in poorer outcomes in BD, when to intervene, and whether alcohol use affects mood symptoms
even at subclinical levels. This understanding is critical given that (1) general clinical practices function
on the perspective that alcohol is primarily used as self-medication with little longitudinal research
to support this viewpoint and (2) extant research has largely focused on co-occurrence of AUD only.
The objective of the present study was to characterize the longitudinal patterns of alcohol use
in one of the largest, ongoing cohort studies of BD, the Prechter Longitudinal Study of Bipolar
Disorder (PLS-BD),21,22 and examine the temporal associations among alcohol use, mood, anxiety,
and functioning over time. We hypothesized that (1) greater alcohol use is associated with more
manic or hypomanic and depressive symptoms, (2) within-person increases in alcohol use are
associated with an increase in manic or hypomanic and depressive symptoms at later time points, (3)
within-person worsening of mood is associated with future increase in alcohol use, and (4) within-
person increases in alcohol use are associated with worse functioning. Given the limited research on
the association between alcohol use and anxiety in BD, we conducted exploratory analyses to
examine these bidirectional associations.

Methods
Participants
Participants for this cohort study were drawn from the PLS-BD, which recruits via advertisements,
psychiatric clinics, mental health centers, and community outreach events across Michigan.21,22 The
University of Michigan Institutional Review Board (IRB) approved the PLS-BD. All participants
provided written informed consent and receive an annual stipend for participation. There were no
additional incentives offered for participation in this specific secondary data analysis. The IRB
approval of PLS-BD applies to the present study, and all participants agreed to the use of their data

JAMA Network Open. 2024;7(6):e2415295. doi:10.1001/jamanetworkopen.2024.15295 (Reprinted) June 7, 2024 2/14

Downloaded from jamanetwork.com by guest on 06/18/2024

 JAMA Network Open | Psychiatry Interplay Between Alcohol Use, Mood, and Functioning in Bipolar Spectrum Disorders

in future secondary data analysis. We followed the Strengthening the Reporting of Observational
Studies in Epidemiology (STROBE) reporting guideline.
The complete PLS-BD cohort currently consists of individuals enrolled for a median (IQR) of 9
(0-16) years. Enrollment, which started in February 2006, is rolling. The present study analyzed data
collected from February 2006 to April 2022. Exclusion criteria were neurological disease and
inability to interview without being intoxicated on alcohol or substances. Selected participants were
those with a diagnosis of BDI or BDII who had been in the study for at least 5 years (eFigure 1 in
Supplement 1). Healthy controls and those with other psychiatric diagnoses were not included in the
present study. Self-reported race and ethnicity data were collected to enhance the generalizability
of the PLS-BD findings to multiple identities. Diagnosis was assessed using the Diagnostic Interview
for Genetic Studies.23 A team of at least 2 doctoral-level psychologists or psychiatrists (including
S.H.S., M.G.M., I.F.T.) confirmed the diagnoses using Diagnostic and Statistical Manual of Mental
Disorders (Fourth Edition, Text Revision) criteria and available medical and treatment history.

Materials and Procedures

Comprehensive information on the study procedures for participants in the PLS-BD is available
elsewhere.21,22 All self-reported measures for the PLS-BD were collected digitally using a secure,
web-based research study platform (REDCap; Vanderbilt University).24,25 Materials and procedures
for the current investigation are detailed herein.
Alcohol consumption, drinking behavior, and alcohol-related problems were measured with the
Alcohol Use Disorders Identification Test (AUDIT)26 at baseline and every 6 months. The AUDIT score
range is from 0 to 40, with 8 or higher indicating AUD is highly probable; 8 to 14 indicating hazardous
or harmful drinking; and 15 to 40 indicating severe drinking or dependence. Internal consistency,
calculated using Cronbach α, was excellent (α = .90). Participants completed 59% of delivered
AUDIT assessments.
The Life Functioning Questionnaire (LFQ),27 developed to assess life functioning in individuals
with BD, was administered at baseline and every 2 months. The LFQ measures functioning over the
past month in 4 domains: leisure time with friends (LFQ friend); leisure time with family (LFQ family);
duties at work, school, or activity center (LFQ work); and duties at home (LFQ home). Items in the
questionnaire are rated using a Likert scale: 0 (does not spend a substantial amount of time in
domain), 1 (no problems), 2 (mild problems), 3 (moderate problems), and 4 (severe problems). These
scores are averaged across each of the 4 domains. Internal consistency for each LFQ subscale was
good (α = .77-.84). Participants completed 57% of delivered LFQ assessments.
The presence and severity of self-reported depressive symptoms over the prior 2 weeks, manic
or hypomanic symptoms over the prior week, and anxiety symptoms over the prior 2 weeks were
measured using the 9-Item Patient Health Questionnaire (PHQ-9; score range: 0-27, with 5-9
indicating mild, 10-14 indicating moderate, 15-19 indicating moderately severe, and ⱖ20 indicating
severe),28 the Altman Self-Rating Mania Scale (ASRM; score range: 0-20, with ⱖ6 indicating high
probability of hypomania),29 and the 7-item Generalized Anxiety Disorder assessment scale (GAD-7;
score range: 0-21, with 5-9 indicating mild, 10-14 indicating moderate, and 15-21 indicating severe).30
The internal consistencies of these instruments were excellent (α = .90) for PHQ-9, good (α = .82)
for ASRM, and excellent (α = .92) for GAD-7. The PHQ-9, ASRM, and GAD-7 were administered at
baseline and every 2 months, and their respective completion rates were 58%, 58%, and 41%.

Statistical Analysis
Summary statistics were estimated using the psych package in RStudio, V.2022.02.0+443
(RStudio).31 To model the temporal dynamics among alcohol use, mood, and functioning, we used
Dynamic Structural Equation Modeling (DSEM)32 to fit multivariable, multilevel first-order vector
autoregressive models in MPlus, version 8.1 (MPlus). A strength of DSEM is its ability to handle
missingness in longitudinal data: DSEM uses the Kalman filter approach, which estimates a missing
observation at timet based on its lagged observation.33 Details regarding DSEM are provided in

JAMA Network Open. 2024;7(6):e2415295. doi:10.1001/jamanetworkopen.2024.15295 (Reprinted) June 7, 2024 3/14

Downloaded from jamanetwork.com by guest on 06/18/2024

 JAMA Network Open | Psychiatry Interplay Between Alcohol Use, Mood, and Functioning in Bipolar Spectrum Disorders

eAppendix in Supplement 1. Seven unconditional models were fit for each measure (AUDIT with
PHQ-9, ASRM, and GAD-7; 4 LFQ domains). These models yielded 8 parameters (2 means,
autocorrelations, cross-lagged regressions, and variances) that were estimated simultaneously and
allowed to vary by individual. For example, for the PHQ-9 and AUDIT model, each individual had the
following 8 parameter scores: (1) μPHQ-9, mean PHQ-9 over time; (2) μAUDIT, mean AUDIT over time;
(3) ␾PP, mean association between PHQ-9 at timet and PHQ-9 6 months later t +1; (4) ␾AA, mean
association between AUDIT at timet and AUDIT 6 months later t +1; (5) ␾AUDIT→PHQ-9, mean
association between AUDIT at timet and PHQ-9 6 months later t +1; (6) ␾PHQ-9→AUDIT, mean
association between PHQ-9 at timet and AUDIT 6 months latert +1; (7) log(πPHQ-9), within-person
variability in PHQ-9 over time; and (8) log(πAUDIT), within-person variability in AUDIT over time. All
other models had the following parameters: ␾MM, autocorrelation of ASRM; ␾GG, autocorrelation of
GAD-7; ␾FF, autocorrelation of LFQ family (in LFQ family model); ␾FF, autocorrelation of LFQ friend
(in LFQ friend model); ␾WW, autocorrelation of LFQ work; and ␾HH, autocorrelation of LFQ home.
Conditional multivariable-multivariate models were then run to examine whether diagnostic
status (0 = BDI, 1 = BDII), sex (0 = female, 1 = male), and time-varying covariates (age; use of mood
stabilizers, antipsychotics, antidepressants, and benzodiazepines: 0 = no, 1 = yes) were associated
with each of the 8 parameters. Results were evaluated based on the 95% credibility interval (CrI);
CrIs that did not contain 0 were considered to be credible differences or associations.

Results
This study included 584 participants (386 females (66.1%), 197 males (33.7%); mean [SD] age, 40
[13.6] years; 5 individuals self-reported as Asian [0.9%], 39 as Black or African American [6.7%], 21 as
Hispanic or Latino [3.6%], 3 as Native American or Alaskan Native [0.5%], 511 as White [87.5%], and
15 as being of more than 1 race and ethnicity [2.6%], with 11 individuals not disclosing [1.9%]). Among
these participants, 445 (76.2%) had a diagnosis of BDI and 139 (23.8%) had BDII (eFigure 1 in
Supplement 1). All summary statistics (eTable 1 in Supplement 1), between-person and within-person
zero-order correlations (eFigure 2 in Supplement 1), model fit indices (eTable 2 in Supplement 1), and
unconditional model results (eTables 3-9 in Supplement 1) are provided in Supplement 1.
Demographic statistics for the entire cohort are available elsewhere.22 Herein we report the results
from the conditional models, including planned covariates.

Longitudinal Patterns of Alcohol Use

At baseline, 246 participants (42.1%) had a lifetime AUD. Across all models, alcohol use was highly
variable over time; however, individuals differed in the intensity of this variability (Figure 1). The
autocorrelation of alcohol use (ie, correlation of increase in problematic drinking at 1 time point with
an increase at the next time point) was more pronounced in BDII than BDI (β = 0.10; 95% CrI, 0.03-
0.19) and less pronounced for individuals taking benzodiazepines compared with individuals not
taking benzodiazepines (β = –0.16; 95% CrI, –0.28 to –0.03) (Table 1). Lower variability in alcohol use
was associated with older age (β = –0.14; 95% CrI, –0.19 to –0.07), antipsychotic use (β = –0.13; 95%
CrI, –0.22 to –0.03), and benzodiazepine use (β = –0.14; 95% CrI, –0.23 to –0.05), whereas higher
variability was associated with antidepressant use (β = 0.10; 95% CrI, 0.01-0.20). Antipsychotic use
was associated with lower mean alcohol use (β = –0.18; 95% CrI, –0.30 to –0.08).

Affective Symptoms and Alcohol Use

A person reporting alcohol use above their own mean amount tended to experience more depressive
symptoms at the next time point, but increased depressive symptoms were not associated with
greater subsequent alcohol use (β = 0.04; 95% CrI, 0.01-0.07) (Figure 2). This cross-lagged
association was not as pronounced for those taking antipsychotics compared with those not taking
antipsychotics (β = –0.21; 95% CrI, –0.39 to –0.02) (Table 1). Previous depression symptoms were
associated with 11.0% of the variance in alcohol use after 6 months, whereas alcohol use was

JAMA Network Open. 2024;7(6):e2415295. doi:10.1001/jamanetworkopen.2024.15295 (Reprinted) June 7, 2024 4/14

Downloaded from jamanetwork.com by guest on 06/18/2024

 JAMA Network Open | Psychiatry Interplay Between Alcohol Use, Mood, and Functioning in Bipolar Spectrum Disorders

associated with 34.0% of the variance in depression symptoms after 6 months. Regarding mania or
hypomania, alcohol use exceeding one’s own mean amount was associated with an increase in manic
or hypomanic symptoms at the next time point, but not vice versa (β = 0.04; 95% CrI, 0.01-0.07])
(Figure 2). Those with BDII showed a more pronounced cross-lagged association between alcohol
use and manic or hypomanic symptoms compared with those with BDI (β = 0.16; 95% CrI,
0.02-0.30) (Table 1). Manic or hypomanic symptoms were associated with 11.0% of the variance in
alcohol use after 6 months, whereas alcohol use was associated with 18.0% of the variance in manic
or hypomanic symptoms after 6 months. Regarding anxiety, no within-person associations emerged
with alcohol use, regardless of covariates (eFigure 3 in Supplement 1; Table 1).

Functioning and Alcohol Use

A person reporting alcohol use above their own mean amount tended to experience a decrease in
work functioning at the next time point, but not vice versa (β = 0.03; 95% CrI, 0.00-0.06)
(Figure 3). This association was more pronounced in BDII than BDI (β = 0.26; 95% CrI, 0.06-0.45)
and for individuals using mood stabilizers (β = 0.38; 95% CrI, 0.03-0.62), and it was less pronounced
for individuals using antidepressants (β = −0.29; 95% CrI, −0.51 to −0.01) (Table 2). Given the large
and near-0 95% CrI for the latter finding, caution should be exercised about overinterpreting the
result. Work functioning was associated with 17.0% of the variance in alcohol use after 6 months,
whereas alcohol use was associated with 40.0% of the variance in work functioning after 6 months.

Figure 1. Longitudinal Patterns of Alcohol Use Disorder

A AUDIT scores for patients with BDI and BDII B AUDIT scores for patients with scores deemed low risk (<8)
or harmful (≥8) at baseline
15 15
Diagnosis Baseline risk
BDI Harmful
BDII Low
10 10
AUDIT score

AUDIT score

5 5

0 0
0 10 20 30 40 0 10 20 30 40
Time, observation No. Time, observation No.

C AUDIT scores for patients with or without antipsychotic medication D AUDIT scores for patients with or without benzodiazepine medication

15 15
Antipsychotic use Benzodiazepine use
No No
Yes Yes
10 10
AUDIT score

AUDIT score

5 5

0 0
0 10 20 30 40 0 10 20 30 40
Time, observation No. Time, observation No.

Shaded regions reflect the SE around the mean. AUDIT indicates Alcohol Use Disorders indicating hazardous or harmful drinking; and 15 to 40 indicating severe drinking or
Identification Test; BDI, bipolar disorder type I; BDII, bipolar disorder type II. The AUDIT dependence.
score range is from 0 to 40, with 8 or higher indicating AUD is highly probable; 8 to 14

JAMA Network Open. 2024;7(6):e2415295. doi:10.1001/jamanetworkopen.2024.15295 (Reprinted) June 7, 2024 5/14

Downloaded from jamanetwork.com by guest on 06/18/2024

 Table 1. Conditional Dynamic Structural Equation Modeling (DSEM) for Affective Measuresa

Between-person independent variables, estimated β (95% CrI)

Estimated mean,
Variablesb β (95% CrI) Age Sex BD diagnosis Mood stabilizer use Antipsychotic use Antidepressant use Benzodiazepine use
Depression
μPHQ-9 8.15 (6.61 to 9.72)c −0.09 (−0.16 to −0.02)c −0.02 (−0.09 to 0.04) 0.08 (0.01 to 0.15)c −0.16 (−0.26 to −0.04)c 0.03 (−0.09 to 0.12) 0.14 (0.04 to 0.25)c 0.13 (0.02 to 0.23)c
c c c
μAUDIT 2.65 (1.98 to 3.33) −0.12 (−0.19 to −0.05) −0.02 (−0.09 to 0.05) 0.01 (−0.07 to 0.09) 0.04 (−0.07 to 0.15) −0.18 (−0.30 to −0.08) 0.06 (−0.06 to 0.16) −0.11 (−0.21 to 0.00)
ϕPP 0.29 (0.12 to 0.46)c 0.02 (−0.10 to 0.12) 0.08 (−0.04 to 0.20) −0.09 (−0.20 to 0.03) −0.21 (−0.38 to −0.04)c 0.10 (−0.08 to 0.25) 0.01 (−0.16 to 0.18) 0.03 (−0.14 to 0.18)
ϕAA 0.51 (0.33 to 0.68)c −0.13 (−0.22 to −0.05)c −0.06 (−0.16 to 0.03) 0.10 (0.03 to 0.19)c 0.00 (−0.14 to 0.15) 0.03 (−0.10 to 0.17) −0.03 (−0.17 to 0.10) −0.16 (−0.28 to −0.03)c
ϕAUDIT→PHQ-9 −0.07 (−0.27 to 0.15) 0.10 (−0.12 to 0.36) −0.03 (−0.26 to 0.20) 0.12 (−0.13 to 0.43) 0.22 (−0.14 to 0.58) 0.06 (−0.29 to 0.54) −0.32 (−0.62 to 0.01) 0.15 (−0.25 to 0.48)
ϕPHQ-9→AUDIT 0.01 (−0.02 to 0.03) −0.00 (−0.13 to 0.12) 0.01 (−0.12 to 0.12) −0.11 (−0.24 to 0.02) 0.12 (−0.08 to 0.29) −0.21 (−0.39 to −0.02)c −0.00 (−0.19 to 0.19) −0.01 (−0.17 to 0.17)
JAMA Network Open | Psychiatry

log(πPHQ-9) 2.76 (2.24 to 3.28)c −0.05 (−0.13 to 0.02) −0.07 (−0.14 to −0.00)c 0.06 (−0.01 to 0.13) −0.16 (−0.28 to −0.05)c 0.04 (−0.06 to 0.15) 0.15 (0.03 to 0.25)c 0.12 (0.02 to 0.23)c
c c c c
log(πAUDIT) 1.30 (0.29 to 2.29) −0.14 (−0.19 to −0.07) −0.02 (−0.08 to 0.04) −0.01 (−0.07 to 0.06) 0.04 (−0.06 to 0.14) −0.13 (−0.22 to −0.03) 0.10 (0.01 to 0.20) −0.14 (−0.23 to −0.05)c
Mania

Downloaded from jamanetwork.com by guest on 06/18/2024

μASRM 1.35 (0.66 to 2.04)c 0.06 (−0.01 to 0.13) 0.15 (0.08 to 0.23)c −0.01 (−0.09 to 0.07) 0.00 (−0.12 to 0.11) −0.05 (−0.15 to 0.07) 0.05 (−0.05 to 0.16) 0.19 (0.08 to 0.30)c
c
ϕMM 0.21 (0.03 to 0.39) 0.09 (−0.02 to 0.19) 0.01 (−0.10 to 0.12) −0.10 (−0.19 to 0.01) −0.03 (−0.19 to 0.13) 0.04 (−0.12 to 0.19) −0.03 (−0.19 to 0.11) −0.12 (−0.26 to 0.02)
ϕAUDIT→ASRM 0.08 (−0.09 to 0.23) −0.05 (−0.23 to 0.13) −0.05 (−0.21 to 0.12) 0.16 (0.02 to 0.30)c −0.12 (−0.38 to 0.12) 0.14 (−0.11 to 0.35) −0.21 (−0.44 to 0.02) 0.11 (−0.12 to 0.32)
ϕASRM→AUDIT −0.01 (−0.05 to 0.02) 0.08 (−0.08 to 0.22) −0.06 (0.19 to −0.22) −0.00 (−0.17 to 0.16) 0.15 (−0.08 to 0.38) −0.06 (−0.30 to 0.18) −0.02 (−0.25 to 0.21) −0.11 (−0.31 to 0.09)
log(πASRM) 1.20 (0.00.53 to 1.88)c −0.00.01 (−0.07 to 0.06) 0.00 (−0.07 to 0.07) 0.01 (−0.05 to 0.07) −0.05 (−0.16 to 0.06) 0.05 (−0.05 to 0.14) −0.05 (−0.15 to 0.06) 0.14 (0.03 to 0.24)c
Anxiety
μGAD-7 8.03 (6.12 to 9.90)c −0.16 (−0.23 to −0.08)c −0.06 (−0.14 to 0.02) 0.08 (0.01 to 0.15)c −0.14 (−0.25 to −0.03)c 0.04 (−0.08 to 0.15) 0.07 (−0.05 to 0.18) 0.29 (0.19 to 0.39)c
ϕGG 0.19 (−0.08 to 0.44) 0.03 (−0.10 to 0.15) 0.06 (−0.08 to 0.19) 0.02 (−0.11 to 0.14) −0.05 (−0.24 to 0.11) 0.04 (−0.13 to 0.20) 0.09 (−0.11 to 0.27) −0.04 (−0.22 to 0.12)
ϕAUDIT→GAD-7 0.19 (−0.02 to 0.46) −0.15 (−0.40 to 0.11) 0.00 (−0.34 to 0.28) −0.02 (−0.29 to 0.23) −0.24 (−0.69 to 0.35) 0.07 (−0.35 to 0.53) 0.06 (−0.35 to 0.52) −0.15 (−0.60 to 0.26)
ϕGAD-7→AUDIT −0.02 (−0.05 to 0.02) 0.11 (−0.06 to 0.26) 0.03 (−0.14 to 0.20) −0.01 (−0.17 to 0.15) 0.06 (−0.19 to 0.27) −0.13 (−0.38 to 0.11) −0.03 (−0.26 to 0.21) −0.03 (−0.23 to 0.17)
log(πGAD-7) 3.49 (2.74 to 4.19)c −0.15 (−0.22 to −0.08)c −0.08 (−0.16 to −0.01)c −0.01 (−0.08 to 0.06) −0.16 (−0.27 to −0.05)c −0.07 (−0.18 to 0.05) 0.08 (−0.03 to 0.18) 0.17 (0.06 to 0.27)c

JAMA Network Open. 2024;7(6):e2415295. doi:10.1001/jamanetworkopen.2024.15295 (Reprinted)

Abbreviations: ASRM, Altman Self-Rating Mania Scale; BD, bipolar disorder; CrI, credibility interval.
a
Estimated means ( 95% CrIs) represent the mean of that DSEM parameter across the sample. Estimates β (95% CrIs) include standardized estimates and 95% CrIs. For example, patients using anticonvulsant mood stabilizers had
a mean depression score that was 0.16 (95% CrI, −0.26 to −0.04) SDs below the estimated mean (8.15; 95% CrI, 6.61-9.72). Note that μAUDIT, ␾AA, and log(πAUDIT) are not repeated in each model to reduce redundancy in the table.
b
See the Methods section for details on the parameters for the variables.
c
The 95% CrI does not include 0, suggesting that the association is credible.

June 7, 2024
Interplay Between Alcohol Use, Mood, and Functioning in Bipolar Spectrum Disorders

6/14
 JAMA Network Open | Psychiatry Interplay Between Alcohol Use, Mood, and Functioning in Bipolar Spectrum Disorders

No within-person associations emerged between AUDIT scores and the family, friend, or home
functioning domains (eFigure 4 in Supplement 1).

Discussion
Consistent with our hypotheses, more problematic drinking was associated with depression and
mania or hypomania later. Increased depression and mania or hypomania was not associated with
greater subsequent alcohol use. Furthermore, greater problematic alcohol use was associated with
worse functioning at work but was not associated with functioning in other domains later. Taken
together, these results highlight the role that alcohol use may play in ongoing mood instability and
functional impairment in BD.
The findings regarding the temporal association of depression and mania or hypomania with
alcohol use are inconsistent with the self-medication hypothesis of addiction.34 The self-medication

Figure 2. Conditional Dynamic Structural Equation Modeling (DSEM) for Affective Measures

A Within-person parameters estimated in the B A patient’s cross-lagged regression between AUDIT and PHQ-9
conditional DSEM with PHQ-9 and AUDIT measures (left) and between PHQ-9 and AUDIT (right) at the next time point
30 40

30
φA
φAA
AUDITt-1 UD
AUDITt AUDIT 20
IT
→
PH
Q– log(πAUDIT)

AUDITt
9
PHQt

T 20
DI
AU
– 9→
PHQ-9t-1 HQ PHQ-9t PHQ-9
φP 10
φPP
log(πPHQ–9) 10

0 0
0 10 20 30 40 0 10 20 30
AUDITt-1 PHQt-1

C Within-person parameters estimated in the D A patient’s cross-lagged regression between AUDIT and ASRM
conditional DSEM with ASRM and AUDIT measures (left) and between ASRM and AUDIT (right) at the next time point
20 40

AAφ
AUDITt-1 φA AUDITt AUDIT 15 30
UD
IT
→
AS
RM log(πAUDIT)
AUDITt
ASRMt

T
DI 10 20
AU
→
SRM
ASRMt-1 φA ASRMt ASRM
φMM
log(πASRM) 5 10

0 0
0 10 20 30 40 0 5 10 15 20
AUDITt-1 ASRMt-1

A and C, Dashed lines represent noncredible differences (95% CrI contains 0), and solid
lines represent credible differences (95% CrI does not include 0). t indicates time; t-1,
time – 1 observation; ␾AA, autocorrelation of Alcohol Use Disorders Identification Test
(AUDIT); ␾ASRM→AUDIT, cross-lagged association between Altman Self-Rating Mania
Scale (ASRM) and AUDIT at the next time point; ␾AUDIT→ASRM, cross-lagged association
between AUDIT and ASRM at the next time point; ␾AUDIT→PHQ-9, cross-lagged
association between AUDIT and 9-Item Patient Health Questionnaire (PHQ-9) at the
next time point; ␾MM, autocorrelation of ASRM; ␾PHQ-9→AUDIT, cross-lagged association
between PHQ-9 and AUDIT at the next time point; ␾PP, autocorrelation of PHQ-9;
log(πASRM), within-person variability in ASRM; log(πAUDIT), within-person variability in
AUDIT; log(πPHQ-9), within-person variability in PHQ-9. B and C, Each tan line represents
a participant in the study. The blue solid line represents the group mean with SE around
the mean.

JAMA Network Open. 2024;7(6):e2415295. doi:10.1001/jamanetworkopen.2024.15295 (Reprinted) June 7, 2024 7/14

Downloaded from jamanetwork.com by guest on 06/18/2024

 JAMA Network Open | Psychiatry Interplay Between Alcohol Use, Mood, and Functioning in Bipolar Spectrum Disorders

hypothesis posits that addiction is associated with use of substances to mitigate distressing
symptoms. Thus, in individuals with psychiatric disorders, the behavioral motivator of substance use
may be to reduce or increase specific emotional states.35 Alcohol, given its fast action and temporary
relief as a central nervous system depressant, may be particularly prone to such use. By this logic,
we would have expected that increased affective symptoms would have been associated with
greater alcohol use over time. However, we found that the opposite (ie, increased problematic
alcohol use) was associated with subsequent worsening of depressive and manic or hypomanic
symptoms. Notably, there were no significant individual differences in this random effect, indicating
that this temporal pattern was consistent across individuals.
We also found diagnostic differences in alcohol use. Individuals with BDII exhibited higher
autocorrelation in AUDIT scores, indicating that greater alcohol use in this group was more likely to
persist over time. These individuals also exhibited more pronounced worsening of workplace
function following a period of increased problematic alcohol use. Increased manic or hypomanic
symptoms after increased alcohol use were more pronounced in BDII than in BDI. Prior literature has
linked greater alcohol use to mania or hypomania in BD in combined samples,12,36,37 but these
studies did not examine whether these associations differed across subtypes. By conducting this
delineated analysis, we provide insights into subtype-specific dynamics of alcohol use and mania or
hypomania. However, the diagnostic findings seemingly contradict earlier studies that linked
co-occurring alcohol use with BDI to a worse illness course than with BDII.9,12 While these previous
studies assessed illness course in different ways, the present study of autocorrelation of alcohol use
and functioning focused on previously unexplored domains.
Medication use emerged as a factor associated with the temporal dynamics of alcohol use.
Individuals using antipsychotics reported lower mean AUDIT scores, less variable alcohol use, and a
lower likelihood of depression playing a role in future alcohol use. Individuals using mood stabilizers
exhibited more pronounced worsening of work functioning after a period of greater problematic
alcohol use. Benzodiazepine use was associated with a lower autocorrelation in alcohol use but
higher variability in depression and more manic or hypomanic symptoms. In contrast, individuals
using antidepressants tended to have greater variability in alcohol use. Notably, these results should
not be interpreted as causal as they were simply associations. It is plausible that persons with
co-occurring BD and alcohol use are more likely to be prescribed antipsychotics or that individuals

Figure 3. Conditional Dynamic Structural Equation Modeling (DSEM) for Life Functioning

A Within-person parameters estimated in the B A patient’s cross-lagged regression between AUDIT and LFQ work
conditional DSEM with LFQ work and AUDIT measures (left) and between LFQ work and AUDIT (right) at the next time point
4 40

φAA 3 30
AUDITt-1 φA AUDITt AUDIT
UD
IT
→
W log(πAUDIT)
or
AUDITt
WORKt

k
2 20
D IT
AU
k→
or WORKt
WORKt-1 φW
WORK
φWW 1 10
log(πWork)

0 0
0 10 20 30 40 0 1 2 3 4
AUDITt-1 WORKt-1

A, Dashed lines represent noncredible differences (95% credible interval [CrI] contains
0), and solid lines represent credible differences (95% CrI does not include 0).
t indicates time; t-1, time – 1 observation; ␾AA, autocorrelation of Alcohol Use Disorders
Identification Test (AUDIT); ␾AUDIT→Work, cross-lagged association between AUDIT and
Life Functioning Questionnaire (LFQ) work at the next time point; ␾Work→AUDIT, cross-
lagged association between LFQ work and AUDIT at the next time point; log(πAUDIT)
, within-person variability in AUDIT; log(πWork), within-person variability of LFQ work;
␾WW, autocorrelation of LFQ work. B, Each tan line represents a participant in the study.
The blue solid line represents the group mean with SE around the mean.

JAMA Network Open. 2024;7(6):e2415295. doi:10.1001/jamanetworkopen.2024.15295 (Reprinted) June 7, 2024 8/14

Downloaded from jamanetwork.com by guest on 06/18/2024

 Table 2. Conditional Dynamic Structural Equation Modeling (DSEM) for Functioning Measuresa

Between-person independent variables, estimated β (95% CrI)

Estimated mean,
Variablesb β (95% CrI) Age Sex BD diagnosis Mood stabilizer use Antipsychotic use Antidepressant use Benzodiazepine use
Life Functioning Questionnaire family
μFam 1.55 (1.33 to 1.81)c 0.01 (−0.08 to 0.08) −0.03 (−0.11 to 0.05) −0.02 (−0.10 to 0.06) −0.14 (−0.26 to −0.01)c −0.01 (−0.13 to 0.11) −0.06 (−0.18 to 0.07) 0.12 (−0.01 to 0.24)
c c
μAUDIT 2.32 (1.58 to 3.06) −0.09 (−0.16 to −0.02) −0.02 (−0.10 to 0.05) 0.01 (−0.06 to 0.09) 0.08 (−0.05 to 0.19) −0.20 (−0.31 to −0.07)c 0.04 (−0.08 to 0.16) −0.10 (−0.20 to 02)
ϕFF 0.39 (0.17 to 0.59)c −0.09 (−0.20 to 0.03) 0.05 (−0.07 to 0.16) −0.09 (−0.20 to 0.02) −0.13 (−0.30 to 0.05) 0.13 (−0.04 to 0.30) 0.01 (−0.17 to 0.18) 0.01 (−0.17 to 0.18)
ϕAA 0.40 (0.21 to 0.61)c −0.09 (−0.18 to 0.01) −0.06 (−0.15 to 0.03) 0.14 (0.04 to 0.23)c 0.01 (−0.15 to 0.16) 0.03 (−0.12 to 0.18) −0.05 (−0.20 to 0.09) −0.21 (−0.34 to −0.06)c
ϕAUDIT→Fam 0.01 (−0.04 to 0.06) −0.10 (−0.33 to 0.11) 0.05 (−0.14 to 0.20) 0.04 (−0.18 to 0.26) 0.02 (−0.34 to 0.35) 0.23 (−0.07 to 0.49) −0.15 (−0.43 to 0.14) 0.07 (−0.23 to 0.36)
ϕFam→AUDIT −0.00 (−0.05 to 0.05) −0.00 (−0.24 to 0.22) 0.03 (−0.20 to 0.28) 0.05 (−0.20 to 0.28) 0.02 (−0.31 to 0.33) 0.06 (−0.32 to 0.39) −0.09 (−0.41 to 0.27) 0.02 (−0.30 to 0.33)
JAMA Network Open | Psychiatry

log(πFam) −1.37 (−2.02 to −0.71)c 0.04 (−0.04 to 0.10) −0.05 (−0.13 to 0.01) 0.00 (−0.07 to 0.07) −0.15 (−0.26 to −0.04)c 0.03 (−0.08 to 0.14) 0.04 (−0.07 to 0.15) 0.10 (−0.01 to 0.21)
log(πAUDIT) 0.61 (−0.47 to 1.75) −0.15 (−0.21 to −0.09)c −0.00 (−0.06 to 0.06) 0.02 (−0.04 to 0.08) 0.11 (0.00 to 0.21)c −0.11 (−0.22 to −0.01)c 0.09 (−0.02 to 0.20) −0.15 (−0.24 to −0.04)c
Life Functioning Questionnaire friend

Downloaded from jamanetwork.com by guest on 06/18/2024

μFnd 1.25 (1.04 to 1.47)c −0.06 (−0.14 to 0.02) 0.02 (−0.06 to 0.10) 0.05 (−0.03 to 0.13) 0.00 (−0.13 to 0.13) 0.08 (−0.05 to 0.21) 0.09 (−0.04 to 0.22) 0.02 (−0.12 to 0.14)
ϕFF 0.34 (0.11 to 0.54)c −0.10 (−0.20 to 0.02) 0.07 (−0.05 to 0.17) −0.08 (−0.19 to 0.04) −0.08 (−0.26 to 0.02) 0.19 (0.03 to 0.36)c 0.06 (−0.12 to 0.21) 0.07 (−0.1 to 0.24)
ϕAUDIT→Fnd 0.01 (−0.03 to 0.05) 0.01 (−0.20 to 0.17) −0.08 (−0.27 to 0.08) 0.04 (−0.15 to 0.22) 0.10 (−0.24 to 0.44) −0.03 (−0.33 to 0.29) −0.11 (−0.36 to 0.17) −0.28 (−0.56 to 0.03)
ϕFnd→AUDIT −0.00 (−0.05 to 0.05) 0.00 (−0.25 to 0.23) −0.03 (−0.26 to 0.22) 0.01 (−0.24 to 0.24) 0.03 (−0.29 to 0.35) −0.02 (−0.41 to 0.32) 0.06 (−0.28 to 0.42) −0.02 (−0.34 to 0.29)
log(πFnd) −1.68 (−2.35 to −1.05)c 0.01 (−0.06 to 0.07) −0.03 (−0.10 to 0.04) −0.00 (−0.07 to 0.06) −0.14 (−0.24 to −0.03)c 0.14 (0.03 to 0.24)c 0.09 (−0.02 to 0.19) 0.11 (0.01 to 0.23)c
Life Functioning Questionnaire work
μWork 1.63 (1.38 to 1.90)c 0.03 (−0.05 to 0.10) −0.08 (−0.15 to −0.01)c 0.02 (−0.06 to 0.09) −0.16 (−0.27 to −0.04)c 0.06 (−0.05 to 0.18) 0.23 (0.12 to 0.34)c 0.08 (−0.04 to 0.19)
c c
ϕWW 0.33 (0.13 to 0.54) −0.09 (−0.20 to 0.04) 0.14 (0.01 to 0.25) −0.04 (−0.16 to 0.08) −0.09 (−0.27 to 0.10) 0.03 (−0.15 to 0.19) 0.11 (−0.07 to 0.29) −0.15 (−0.33 to 0.02)
ϕAUDIT→Work −0.04 (−0.07 to −0.00)c 0.11 (−0.11 to 0.29) 0.03 (−0.14 to 0.21) 0.26 (0.06 to 0.45)c 0.38 (0.03 to 0.62)c 0.05 (−0.27 to 0.36) −0.29 (−0.51 to −0.01)c −0.03 (−0.31 to 0.27)
ϕWork→AUDIT −0.00 (−0.05 to 0.05) −0.00 (−0.24 to 0.23) 0.04 (−0.21 to 0.28) 0.03 (−0.21 to 0.27) 0.10 (−0.24 to 0.40) −0.11 (−0.46 to 0.22) −0.03 (−0.35 to 0.32) 0.04 (−0.28 to 0.34)
log(πWork) −1.85 (−2.41 to −1.29)c 0.08 (0.01 to 0.15)c −0.14 (−0.21 to −0.07)c −0.00 (−0.07 to 0.06) −0.08 (−0.19 to 0.04) 0.07 (−0.03 to 0.19) 0.16 (0.05 to 0.26)c 0.01 (−0.10 to 0.13)

JAMA Network Open. 2024;7(6):e2415295. doi:10.1001/jamanetworkopen.2024.15295 (Reprinted)

Life Functioning Questionnaire home
μHome 1.90 (1.67 to 2.17)c −0.35 (−0.42 to −0.28)c −0.04 (−0.10 to 0.03) 0.02 (−0.05 to 0.08) −0.04 (−0.17 to 0.06) −0.05 (−0.16 to 0.05) 0.06 (−0.05 to 0.17) −0.21 (−0.32 to −0.11)c
c
ϕHH 0.23 (0.01 to 0.43) 0.02 (−0.10 to 0.15) 0.04 (−0.08 to 0.15) −0.03 (−0.14 to 0.08) −0.13 (−0.31 to 0.05) 0.12 (−0.05 to 0.28) 0.05 (−0.12 to 0.21) −0.07 (−0.24 to 0.12)
ϕAUDIT→Home 0.00 (−0.03 to 0.04) −0.13 (−0.31 to 0.08) 0.04 (−0.14 to 0.21) 0.06 (−0.17 to 0.27) 0.07 (−0.22 to 0.33) 0.09 (−0.17 to 0.34) 0.09 (−0.18 to 0.35) 0.03 (−0.26 to 0.31)
ϕHome→AUDIT 0.00 (−0.06 to 0.06) −0.05 (−0.29 to 0.19) −0.02 (−0.29 to 0.24) 0.01 (−0.29 to 0.31) 0.05 (−0.37 to 0.41) 0.01 (−0.43 to 0.39) 0.08 (−0.29 to 0.44) −0.04 (−0.41 to 0.33)
log(πHome) 0.05 (−0.73 to 0.87) −0.16 (−0.23 to −0.10)c −0.09 (−0.15 to −0.02)c −0.05 (−0.11 to 0.01) −0.03 (0.13 to 0.08) 0.00 (−0.10 to 0.11) 0.10 (−0.01 to 0.20) −0.15 (−0.26 to −0.05)c

Abbreviations: BD, bipolar disorder; CrI, credibility interval.

a
Estimated means (95% CrIs) represent the mean of that DSEM parameter across the sample. Estimated β (95% CrIs) include standardized estimates and 95% CrIs. For example, patients using antipsychotic mood stabilizers had
a mean Alcohol Use Disorders Identification Test score that was −0.20 (95% CrI, −0.31 to −0.07) SD below the estimated mean (2.32; 95% CrI, 1.58-3.06). Note that μAUDIT, ␾AA, and log(πAUDIT) are not repeated in each model to
reduce redundancy in the table.
b
See the Methods section for details on the parameters for the variables.
c
The 95% CrI does not include 0, suggesting that the association is credible.

June 7, 2024
Interplay Between Alcohol Use, Mood, and Functioning in Bipolar Spectrum Disorders

9/14
 JAMA Network Open | Psychiatry Interplay Between Alcohol Use, Mood, and Functioning in Bipolar Spectrum Disorders

with higher depression and mania or hypomania are more likely to be using benzodiazepines during
periods of higher symptom severity. Nonetheless, these medication-related findings emphasize the
need for careful consideration of medication regimens in managing patients with BD who
drink alcohol.
We did not find within-person or between-person associations between alcohol use and
anxiety. Because this study used a longer timescale, the length of time between each assessment
period may have concealed the temporal associations between a change in anxiety and change in
drinking. For example, alcohol may be used as an acute remedy to anxiety in specific instances,
thereby reducing the likelihood that it would emerge in a model using 6-month increments. Studies
that use in-the-moment methods, such as ecological momentary assessment, might be better suited
for examining the more proximal bidirectional associations between alcohol use and anxiety.38
While ecological momentary assessment could provide more proximal insights, the present
study’s 6-month timescale has strengths. First, it enabled us to examine long-term patterns of
alcohol use (as opposed to individual alcohol binge episodes) and mood (which is more prolonged
than emotion). Furthermore, it allowed us to examine the implications of age for temporal dynamics
of alcohol use in BD, which, on average, becomes less problematic and variable. Nonetheless,
administering the AUDIT more frequently may help dissect the temporal association between
increased alcohol use and mood. Although we found that increased alcohol use was associated with
later mood symptoms, it remains plausible that this alcohol use was preceded by the beginning
stages of a mood episode that the measures did not capture.
These findings hold critical clinical implications. Fluctuations in alcohol use were common in the
sample and were associated with unstable mood and functional impairment at work, regardless of
whether an individual had AUD. Thus, regular monitoring of alcohol use with the AUDIT or similar
tools may be advisable in treating patients with BD. Relatedly, patients with BD may benefit from
reduced alcohol consumption if their treatment goals included improving work functioning and/or
reducing mood symptoms, complemented by other skills to prevent future problematic alcohol use.
Notably, many individuals without AUD reported AUDIT scores that may benefit from intervention
(eg, score of 8-14); thus, interventions focused on alcohol use in BD may be considered for anyone
engaging in problematic alcohol use regardless of diagnostic status.
Because little research has evaluated integrated treatments for alcohol use and BD,6,39 it is
unclear what kind of alcohol use treatment would be most helpful in BD. Future studies are needed
to examine whether abstinence (refraining from any alcohol use) vs harm-reduction methods (self-
moderation and reducing frequency or amount)40 differentially alter mood, functioning, and course
of illness. Follow-up studies could focus on identifying motives or risk factors that may precede
increases in alcohol use. Just-in-time adaptive interventions in combination with passive sensing
technology could be deployed to a person who engages in alcohol risk behaviors. In sum, the present
findings provide multiple avenues for future clinical intervention and research.

Limitations
This study was limited by several aspects of the PLS-BD protocol and overall participant demographic
makeup. First, this study was naturalistic in design, and therefore it is unclear whether specific types
and timing of treatment (eg, medication changes and therapy) changed the longitudinal dynamics of
alcohol use, mood, and functioning. Second, the PLS-BD cohort lacks racial diversity; most
participants are White individuals. Although this demographic makeup mirrors that of the PLS-BD
recruitment catchment area, it does not reflect the demographic characteristics of the US as a whole
and therefore may limit the findings’ generalizability.

Conclusions
The findings of this study suggest that alcohol use, regardless of AUD diagnostic status, may
destabilize the course of BD, as it was associated with mood instability and poorer work functioning

JAMA Network Open. 2024;7(6):e2415295. doi:10.1001/jamanetworkopen.2024.15295 (Reprinted) June 7, 2024 10/14

Downloaded from jamanetwork.com by guest on 06/18/2024

 JAMA Network Open | Psychiatry Interplay Between Alcohol Use, Mood, and Functioning in Bipolar Spectrum Disorders

and did not appear to be a response to mood symptom changes. Baseline measures of the
co-occurrence between BD and AUD diagnoses were not sufficient to capture the dynamic nature of
alcohol use in this patient population. Greater efforts toward the dimensional and longitudinal
assessment and management of alcohol use are necessary and should be integrated into research
and standard treatment for individuals with BD.

ARTICLE INFORMATION
Accepted for Publication: April 5, 2024.
Published: June 7, 2024. doi:10.1001/jamanetworkopen.2024.15295
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2024 Sperry SH
et al. JAMA Network Open.
Corresponding Author: Sarah H. Sperry, PhD, Department of Psychiatry, University of Michigan,
4250 Plymouth Rd, Ann Arbor, MI 48109 (sperrys@med.umich.edu).
Author Affiliations: Department of Psychology, University of Michigan, Ann Arbor (Sperry, Stromberg, Lasagna,
Menkes); Department of Psychiatry, University of Michigan, Ann Arbor (Sperry, Stromberg, Murphy, Lasagna,
McInnis, Menkes, Yocum, Tso); Department of Psychiatry and Behavioral Health, The Ohio State University,
Columbus (Tso).
Author Contributions: Dr. Sperry and Ms. Stromberg had full access to all of the data in the study and take
responsibility for the integrity of the data and the accuracy of the data analysis. Dr. Sperry and Ms. Stromberg
contributed equally to this work and share co–first authorship.
Concept and design: Sperry, Stromberg, Murphy, McInnis, Tso.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Sperry, Stromberg, Murphy, Menkes, Yocum.
Critical review of the manuscript for important intellectual content: Sperry, Stromberg, Murphy, Lasagna,
McInnis, Tso.
Statistical analysis: Sperry, Stromberg, Murphy, Lasagna, Yocum, Tso.
Obtained funding: Sperry, McInnis, Tso.
Administrative, technical, or material support: Sperry, Stromberg, McInnis.
Supervision: Sperry, Stromberg, McInnis, Tso.
Conflict of Interest Disclosures: Dr Sperry reported receiving a philanthropic gift from Prechter Family Charitable
Fund and Richard Tam Foundation and grants from National Institute of Mental Health (NIMH) and Brain and
Behavior Research Foundation during the conduct of the study. Ms Stromberg reported receiving a philanthropic
gift to the Heinz C. Prechter Bipolar Research Program from the Prechter Family Charitable Fund during the
conduct of the study. Dr McInnis reported receiving a philanthropic gift from Prechter Family Charitable Fund and
Richard Tam Foundation during the conduct of the study and grants from Janssen Pharmaceuticals outside the
submitted work. Ms Menkes reported receiving a philanthropic gift from Prechter Family Charitable Fund and
Richard Tam Foundation outside the submitted work. Dr Yocum reported receiving salary support from Prechter
Family Charitable Fund and personal fees from A2IDEA LLC outside the submitted work. Dr Tso reported receiving
grants from NIMH during the conduct of the study. No other disclosures were reported.
Funding/Support: This research was supported by the Prechter Family Charitable Fund and the Richard Tam
Foundation and was based on work supported by Graduate Research Fellowship grant DGE-1841052 from the
National Science Foundation under (Ms Lasagna), One Mind Bipolar Research Award (Dr Tso), NIMH R01MH122491
(Dr Tso), Baszucki Brain Research Fund (Dr Tso), grant K23MH131601 from NIMH (Dr Sperry), grant L30MH127613
from NIMH (Dr Sperry), and Brain and Behavior Research Foundation Young Investigator Award (Dr Sperry).
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection,
management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and
decision to submit the manuscript for publication.
Disclaimer: Any opinion, findings, conclusions, or recommendations expressed herein are those of the authors
and do not necessarily reflect the views of the National Science Foundation.
Data Sharing Statement: See Supplement 2.

JAMA Network Open. 2024;7(6):e2415295. doi:10.1001/jamanetworkopen.2024.15295 (Reprinted) June 7, 2024 11/14

Downloaded from jamanetwork.com by guest on 06/18/2024

 JAMA Network Open | Psychiatry Interplay Between Alcohol Use, Mood, and Functioning in Bipolar Spectrum Disorders

Additional Contributions: With gratitude, we acknowledge the participants in the University of Michigan Prechter
Bipolar Longitudinal Research. The research team of the Prechter Bipolar Research Program performed the
collection and stewardship of the data.

REFERENCES
1. Cardoso BM, Kauer Sant’Anna M, Dias VV, Andreazza AC, Ceresér KM, Kapczinski F. The impact of co-morbid
alcohol use disorder in bipolar patients. Alcohol. 2008;42(6):451-457. doi:10.1016/j.alcohol.2008.05.003
2. Frye MA, Salloum IM. Bipolar disorder and comorbid alcoholism: prevalence rate and treatment considerations.
Bipolar Disord. 2006;8(6):677-685. doi:10.1111/j.1399-5618.2006.00370.x
3. Regier DA, Farmer ME, Rae DS, et al. Comorbidity of mental disorders with alcohol and other drug abuse.
Results from the Epidemiologic Catchment Area (ECA) study. JAMA. 1990;264(19):2511-2518. doi:10.1001/jama.
1990.03450190043026
4. Goldberg JF, Harrow M, Grossman LS. Course and outcome in bipolar affective disorder: a longitudinal
follow-up study. Am J Psychiatry. 1995;152(3):379-384. doi:10.1176/ajp.152.3.379
5. Hasin DS, Stinson FS, Ogburn E, Grant BF. Prevalence, correlates, disability, and comorbidity of DSM-IV alcohol
abuse and dependence in the United States: results from the National Epidemiologic Survey on Alcohol and
Related Conditions. Arch Gen Psychiatry. 2007;64(7):830-842. doi:10.1001/archpsyc.64.7.830
6. Farren CK, Hill KP, Weiss RD. Bipolar disorder and alcohol use disorder: a review. Curr Psychiatry Rep. 2012;14
(6):659-666. doi:10.1007/s11920-012-0320-9
7. Nolen WA, Luckenbaugh DA, Altshuler LL, et al. Correlates of 1-year prospective outcome in bipolar disorder:
results from the Stanley Foundation Bipolar Network. Am J Psychiatry. 2004;161(8):1447-1454. doi:10.1176/appi.
ajp.161.8.1447
8. Tohen M, Waternaux CM, Tsuang MT. Outcome in Mania. A 4-year prospective follow-up of 75 patients utilizing
survival analysis. Arch Gen Psychiatry. 1990;47(12):1106-1111. doi:10.1001/archpsyc.1990.01810240026005
9. Preuss UW, Hesselbrock MN, Hesselbrock VM. A prospective comparison of bipolar I and II subjects with and
without comorbid alcohol dependence from the COGA dataset. Front Psychiatry. 2020;11:522228. doi:10.3389/
fpsyt.2020.522228
10. Sonne SC, Brady KT, Morton WA. Substance abuse and bipolar affective disorder. J Nerv Ment Dis. 1994;182
(6):349-352. doi:10.1097/00005053-199406000-00007
11. Nery FG, Miranda-Scippa A, Nery-Fernandes F, Kapczinski F, Lafer B. Prevalence and clinical correlates of
alcohol use disorders among bipolar disorder patients: results from the Brazilian Bipolar Research Network. Compr
Psychiatry. 2014;55(5):1116-1121. doi:10.1016/j.comppsych.2014.02.006
12. Rakofsky JJ, Dunlop BW. Do alcohol use disorders destabilize the course of bipolar disorder? J Affect Disord.
2013;145(1):1-10. doi:10.1016/j.jad.2012.06.012
13. Jaffee WB, Griffin ML, Gallop R, et al. Depression precipitated by alcohol use in patients with co-occurring
bipolar and substance use disorders. J Clin Psychiatry. 2009;70(2):171-176. doi:10.4088/JCP.08m04011
14. Salloum IM, Cornelius JR, Mezzich JE, Kirisci L. Impact of concurrent alcohol misuse on symptom presentation
of acute mania at initial evaluation. Bipolar Disord. 2002;4(6):418-421. doi:10.1034/j.1399-5618.2002.01194.x
15. Strakowski SM, DelBello MP, Fleck DE, Arndt S. The impact of substance abuse on the course of bipolar
disorder. Biol Psychiatry. 2000;48(6):477-485. doi:10.1016/S0006-3223(00)00900-8
16. Strakowski SM, Keck PE Jr, McElroy SL, et al. Twelve-month outcome after a first hospitalization for affective
psychosis. Arch Gen Psychiatry. 1998;55(1):49-55. doi:10.1001/archpsyc.55.1.49
17. Strakowski SM, Sax KW, McElroy SL, Keck PE Jr, Hawkins JM, West SA. Course of psychiatric and substance
abuse syndromes co-occurring with bipolar disorder after a first psychiatric hospitalization. J Clin Psychiatry.
1998;59(9):465-471. doi:10.4088/JCP.v59n0905
18. González-Pinto A, Alberich S, Barbeito S, et al. Different profile of substance abuse in relation to predominant
polarity in bipolar disorder: the Vitoria long-term follow-up study. J Affect Disord. 2010;124(3):250-255. doi:10.
1016/j.jad.2009.11.005
19. Pavlova B, Perlis RH, Alda M, Uher R. Lifetime prevalence of anxiety disorders in people with bipolar disorder:
a systematic review and meta-analysis. Lancet Psychiatry. 2015;2(8):710-717. doi:10.1016/S2215-0366(15)00112-1
20. Ummels SA, Seldenrijk A, Bos EH, de Graaf R, Batelaan NM, Ten Have M. The bidirectional relationship
between anxiety disorders and alcohol use disorders in adults: findings from a longitudinal population-based
study. J Affect Disord. 2022;314:126-132. doi:10.1016/j.jad.2022.06.091

JAMA Network Open. 2024;7(6):e2415295. doi:10.1001/jamanetworkopen.2024.15295 (Reprinted) June 7, 2024 12/14

Downloaded from jamanetwork.com by guest on 06/18/2024

 JAMA Network Open | Psychiatry Interplay Between Alcohol Use, Mood, and Functioning in Bipolar Spectrum Disorders

21. McInnis MG, Assari S, Kamali M, et al; Prechter Bipolar Clinical Research Collaborative; Prechter Longitudinal
Study of Bipolar Disorder. Cohort profile: the Heinz C. Prechter longitudinal study of bipolar disorder. Int J
Epidemiol. 2018;47(1):28-28n. doi:10.1093/ije/dyx229
22. Yocum AK, Anderau S, Bertram H, et al; Prechter Longitudinal Study of Bipolar Disorder. Cohort profile update:
the Heinz C. Prechter longitudinal study of bipolar disorder. Int J Epidemiol. 2023;52(6):e324-e331. doi:10.1093/
ije/dyad109
23. Nurnberger JI Jr, Blehar MC, Kaufmann CA, et al; NIMH Genetics Initiative. Diagnostic interview for genetic
studies: rationale, unique features, and training. Arch Gen Psychiatry. 1994;51(11):849-859. doi:10.1001/archpsyc.
1994.03950110009002
24. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research Electronic Data Capture (REDCap)–a
metadata-driven methodology and workflow process for providing translational research informatics support.
J Biomed Inform. 2009;42(2):377-381. doi:10.1016/j.jbi.2008.08.010
25. Harris PA, Taylor R, Minor BL, et al; REDCap Consortium. The REDCap consortium: building an international
community of software platform partners. J Biomed Inform. 2019;95:103208. doi:10.1016/j.jbi.2019.103208
26. Saunders JB, Aasland OG, Babor TF, de la Fuente JR, Grant M. Development of the Alcohol Use Disorders
Identification Test (AUDIT): WHO collaborative project on early detection of persons with harmful alcohol
consumption–II. Addiction. 1993;88(6):791-804. doi:10.1111/j.1360-0443.1993.tb02093.x
27. Altshuler L, Mintz J, Leight K. The Life Functioning Questionnaire (LFQ): a brief, gender-neutral scale assessing
functional outcome. Psychiatry Res. 2002;112(2):161-182. doi:10.1016/S0165-1781(02)00180-4
28. Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief depression severity measure. J Gen Intern
Med. 2001;16(9):606-613. doi:10.1046/j.1525-1497.2001.016009606.x
29. Altman EG, Hedeker D, Peterson JL, Davis JM. The Altman Self-Rating Mania Scale. Biol Psychiatry. 1997;42
(10):948-955. doi:10.1016/S0006-3223(96)00548-3
30. Spitzer RL, Kroenke K, Williams JBW, Löwe B. A brief measure for assessing generalized anxiety disorder: the
GAD-7. Arch Intern Med. 2006;166(10):1092-1097. doi:10.1001/archinte.166.10.1092
31. psych: Procedures for psychological, psychometric, and rersonality research. Version 2.2.9. 2022. Accessed
December 3, 2023. https://CRAN.R-project.org/package=psych
32. Asparouhov T, Hamaker EL, Muthén B. Dynamic structural equation models. Struct Equ Modeling. 2018;25(3):
359-388. doi:10.1080/10705511.2017.1406803
33. McNeish D, Hamaker EL. A primer on two-level dynamic structural equation models for intensive longitudinal
data in Mplus. Psychol Methods. 2020;25(5):610-635. doi:10.1037/met0000250
34. Khantzian EJ. The self-medication hypothesis of substance use disorders: a reconsideration and recent
applications. Harv Rev Psychiatry. 1997;4(5):231-244. doi:10.3109/10673229709030550
35. Khantzian EJ. The self-medication hypothesis revisited: the dually diagnosed patient. Prim Psychiatry. 2003;
10(9):47-54.
36. Frye MA, Altshuler LL, McElroy SL, et al. Gender differences in prevalence, risk, and clinical correlates of
alcoholism comorbidity in bipolar disorder. Am J Psychiatry. 2003;160(5):883-889. doi:10.1176/appi.ajp.
160.5.883
37. Goldstein BI, Velyvis VP, Parikh SV. The association between moderate alcohol use and illness severity in
bipolar disorder: a preliminary report. J Clin Psychiatry. 2006;67(1):102-106. doi:10.4088/JCP.v67n0114
38. Wray TB, Merrill JE, Monti PM. Using ecological momentary assessment (EMA) to assess situation-level
predictors of alcohol use and alcohol-related consequences. Alcohol Res. 2014;36(1):19-27.
39. Grunze H, Schaefer M, Scherk H, Born C, Preuss UW. Comorbid bipolar and alcohol use disorder—a therapeutic
challenge. Front Psychiatry. 2021;12:660432. doi:10.3389/fpsyt.2021.660432
40. Marlatt GA, Witkiewitz K. Harm reduction approaches to alcohol use: health promotion, prevention, and
treatment. Addict Behav. 2002;27(6):867-886. doi:10.1016/S0306-4603(02)00294-0

SUPPLEMENT 1.
eFigure 1. Subsample Selection and Sample Size
eAppendix. DSEM Modeling Approach
eTable 1. Demographic and Summary Statistics
eFigure 2. Within and Between Person Spearman Rho Correlations
eTable 2. Potential Scale Reduction and Deviance Information Criterion
eTable 3. Unconditional DSEM Model: PHQ9 and AUDIT
eTable 4. Unconditional DSEM Model: ASRM and AUDIT

JAMA Network Open. 2024;7(6):e2415295. doi:10.1001/jamanetworkopen.2024.15295 (Reprinted) June 7, 2024 13/14

Downloaded from jamanetwork.com by guest on 06/18/2024

 JAMA Network Open | Psychiatry Interplay Between Alcohol Use, Mood, and Functioning in Bipolar Spectrum Disorders

eTable 5. Unconditional DSEM Model: GAD7 and AUDIT

eTable 6. Unconditional DSEM Model: LFQ Family and AUDIT
eTable 7. Unconditional DSEM Model: LFQ Friend and AUDIT
eTable 8. Unconditional DSEM Model: LFQ Family and AUDIT
eTable 9. Unconditional DSEM Model: LFQ Work and AUDIT
eFigure 3. Conditional DSEM Model for GAD7 and AUDIT
eFigure 4. Conditional DSEM Model for LFQ Family, Friend, and Home with AUDIT

SUPPLEMENT 2.
Data Sharing Statement

JAMA Network Open. 2024;7(6):e2415295. doi:10.1001/jamanetworkopen.2024.15295 (Reprinted) June 7, 2024 14/14

Downloaded from jamanetwork.com by guest on 06/18/2024