Clinical Characteristics and Functioning of Adults with Bipolar I Disorder: Evidence from the Mental and Substance Use Disorders Prevalence Study

Abstract

Background:

Knowledge of clinical, treatment and life circumstances of individuals with bipolar I disorder (BP-I) in US households is informed by decades old epidemiological surveys.

Methods:

The Mental and Substance Use Disorders Prevalence Study was conducted October 2020-October 2022. Clinicians administered the Structured Clinical Interview for the DSM-5 diagnosing 12-month prevalence of BP-I and other mental health disorders (MHD) among 4,764 adults aged 18–65 years and collected sociodemographic information. We examined clinical characteristics, differences by sex and age among adults with BP-I, and compared adults with BP-I versus no MHD regarding sociodemographic characteristics, functioning, and substance use disorders (SUDs).

Results:

Prevalence of BP-I in the MDPS was 1.5%. Among those with BP-I, 73.4% had comorbid psychiatric disorders, and 43.4% had comorbid SUDs. Alcohol use disorder was higher in those with BP-I versus no MHD (33.0% vs. 6.3%). Mean Global Assessment of Functioning scores were lower among those with BP-I versus no MHD (53.2 vs. 77.0). Of individuals with BP-I, 64.9% had past-year outpatient, 5.4% inpatient, and 18.7% minimally adequate treatment (≥1 antimanic agent and ≥4 outpatient visits). Individuals with BP-I were less likely to be employed (37.3% vs. 63.0%) and have a family income ≥$20,000 (48.2% vs. 81.9%) versus no MDPS MHD.

Limitations:

The survey response rate was low.

Conclusions:

In this sample, many individuals with BP-I had psychiatric and SUD comorbidities, lived in poverty and had functional impairment. Few received adequate treatment; women and younger individuals were particularly disadvantaged. Early detection and treatment represent substantial opportunities to improve outcomes.

Introduction

Bipolar I disorder (BP-I) is a lifetime condition that is thought to affect approximately 1% of the US population¹. Symptoms and functioning vary over the course of illness, resulting in mood episodes and periods of euthymia as well as periods with impaired and normal functioning. The disorder is associated with premature mortality due to physical comorbidities, comorbid substance use, and suicide^2,3. People with BP-I commonly experience periods of only partial symptom reduction until an effective treatment regimen, which achieves symptom remission, can be established. It often takes up to 10 years from diagnosis to identification of an effective, often complex treatment regimen^4,5. Additionally, among individuals with symptomatic recovery, long-term prospective studies have found that they may still experience new mood episodes and symptoms almost half of the weeks during follow-up despite engaging in effective treatment^6–8.

Ascertaining accurate clinical and epidemiological information about BP-I is complicated by the similarity of symptoms with other disorders including unipolar depression, bipolar II disorder, and substance induced mood episodes or mood episodes due to a general medical condition. Individuals may present with subthreshold symptoms early in the illness trajectory, limiting our understanding of the true magnitude of unmet need. The National Comorbidity Survey Replication (NCS-R), conducted between 2000–2003, provided the most robust and comprehensive epidemiological information on the past-year prevalence of BP-I, its correlates, and service use in a US nationally representative sample. The NCS-R found that although men and women were equally likely to have BP-I, major depressive episodes and suicidal thoughts and behaviors were more common in women than men. Compared to all individuals with bipolar spectrum disorders, those specifically with BP-I were younger, had lower educational attainment, more were previously than currently married, and more were unemployed-disabled than currently employed. Two-thirds of people with BP-I had at least some mental health treatment (67%), but only 25% received what was considered appropriate treatment (i.e., taking at least one mood stabilizer or antipsychotic) in the past year¹. The more recent National Epidemiological Survey on Alcohol and Related Conditions III (NESARC III) from 2012–2013 reported sociodemographic characteristics, functioning and quality of life among individuals with BP-I that were consistent with the NCS-R, and further found that post-traumatic stress disorder (PTSD) was the most common psychiatric comorbidity while alcohol use disorder was the most common substance use comorbidity^2,9.

Limitations of these studies include use of lay interviewers administering fully structured screening interviews (i.e., the NCS-R used the CIDI¹⁰, and NESARC-III used the AUDADIS-5¹¹) and often grouping diagnoses of bipolar spectrum disorders including BP-I, BP-II and subthreshold BP rather than specifically BP-I. Because a considerable number of individuals with BP-I are not in treatment, findings from national clinical samples based on claims data and international samples of individuals who received treatment^12,13 have limited generalizability given their reliance on administrative data and ICD codes to define BP-I.

The Mental Disorders and Substance Use Prevalence Study (MDPS), conducted from October 2020-October 2022, was the first study to use trained clinicians to administer the Structured Clinical Interview of the DSM-5 (SCID-5) to individuals from a national sample of US households. This innovative study identified individuals meeting diagnostic criteria of selected mental health disorders including BP-I and collected detailed information on sociodemographic and clinical characteristics of study participants.

The current study used data from the MDPS household sample to compare clinical characteristics of adults with BP-I with mood episodes in the past year to people without a past-year mental health disorder, and to examine treatments among individuals with BP-I. As effective new psychotherapeutic interventions such as psychoeducation and interpersonal and social rhythm therapy and psychotropic treatments for BP-I have been developed since previous epidemiological surveys were conducted, we hypothesized that the circumstances of individuals with this condition would have improved.

Methods

Data Source and Sampling Design

The overall aim of the MDPS, funded by the U.S. Substance Use and Mental Health Services Administration (SAMHSA), was to obtain unbiased inclusive estimates of the prevalence of mental and substance use disorders, including specifically powering the study to capture the prevalence of SMI for the US population of adults aged 18–65. The MDPS household sample was initially designed as a nationally representative survey to identify current prevalence estimates of mental health and substance use disorders in households in the US, with particular focus on sampling enrichment for serious mental illness (SMI) including BP-I and schizophrenia spectrum disorders. A three-stage probability clustered stratified sampling design was used to select the household roster sample beginning with 1) a roster of all households in the US from 100 primary sampling units (PSUs; i.e., counties), 2) 16 randomly selected secondary sampling units based on census blocks within each PSU, and 3) a random sample of addresses within the 16 secondary sampling units, resulting in a final roster of 234,270 households sent an invitation to participate in the survey. Of these, 25,752 (weighted conditional response rate 17.4%) completed the initial roster and a random sample of up to two adults aged 18–65 years were selected to complete a screening interview resulting in a sample of 41,868. Among those, 29,084 (weighted conditional response rate 67.4%) completed the screening interview which included either the CIDI (Composite International Diagnostic Interview)¹⁰ or the CAT-MH (Computerized Adaptive Test – Mental Health)¹⁴. A stratified random sample was selected to complete the clinical interview based on screening responses so that individuals at elevated risk for mental health disorders would be oversampled; 100% of screen positive cases for psychosis were selected from one strata, a random sample of 80% screen positive cases for other mental or substance use disorders were selected from the second strata, and a random sample of all other individuals from the third strata, with a final sample of 12,906 individuals. The clinical interview was completed by 4,764 respondents (weighted conditional response rate 31.2%, a final response rate of 3.7%).

MDPS household survey weights were designed to reduce nonresponse bias occurring at each stage of the design and to reduce differences in demographics with national estimates based on the American Community Survey (ACS) 2020 5-year estimates. Notably the MDPS sample and ACS-5 difference estimates on several of these variables were on average very small (1.1%). See Ringeisen et al. (2023)¹⁵ and Guyer, et al., (2023)¹⁶ for additional details on MDPS methods. Because of the low overall response rate and need for weighting to address nonresponse bias, we present weighted as well as unweighted adjusted odds ratios.

Unlike prior epidemiologic studies that used fully structured, lay-administered screeners with limited diagnostic accuracy for conditions such as schizophrenia and BP-I, the MDPS used clinicians trained to assess for BP-I using a modified version of the Structured Clinical Interview for DSM-5 (SCID-5),¹⁷ which is a semi-structured diagnostic interview for psychiatric disorders. These clinicians used their clinical judgement to integrate the open-ended Overview with symptoms discussed later in the questionnaire. The SCID-5 allows BP-I disorder to be differentiated from major depressive disorder (MDD) and schizophrenia spectrum disorders (SSD), lending confidence that individuals diagnosed with BP-I did not have MDD or SSD. Interviews were conducted in Spanish and English. The clinicians received extensive training and supervision in the administration of the SCID-5 before and throughout data collection, with calibration between interviewers. Data were collected between October 2020 to October 2022.

Measurement

The clinicians administering the SCID-5 assessed selected mental health and substance use disorders including past year BP-I, MDD, posttraumatic stress (PTSD), obsessive compulsive (OCD), generalized anxiety (GAD), anorexia nervosa, and lifetime and past year SSD. The SCID-5 substance use disorders included past year alcohol use disorder (AUD), cannabis use disorder (CUD), sedative/hypnotic/anxiolytic use disorder, opioid use disorder, and stimulant use disorder. The BP-I category included people meeting DSM-5 criteria specifically for BP-I but were limited to those who experienced a manic episode and/or major depressive episode in the past year. Mood episodes and other symptoms judged to be substance induced or due to another medical condition were not included. Individuals with BP-I who did not experience any mood episodes in the past year were not identified in this sample because people without a mood episode were not asked about lifetime mania. In the following analyses, the no MDPS mental health disorder (MHD) category included individuals who did not meet criteria for past year BP-I, MDD, PTSD, OCD, GAD, anorexia nervosa or past year or lifetime SSD.

Clinical characteristics included major depressive episodes (MDE), manic episodes, and presence of at least one psychotic symptom (not including individuals with only catatonic, avolition, diminished emotional expression or not better explained by substance use or a general medical condition symptoms). Psychiatric comorbidities included GAD, PTSD, OCD, and more than one mental health disorder. Anorexia nervosa is not reported here due to the small number of respondents meeting criteria.

Self-reported lifetime diagnoses received prior to MDPS included bipolar disorder (mania, manic-depression, or bipolar disorder), depression, SSD (schizophrenia or schizoaffective disorder), or other emotional problem. Past-year substance use included alcohol, cannabis, stimulants, opioids, sedatives/hypnotics/anxiolytics, cigarette smoking and vaping. Comorbid substance use disorders (SUDs) included past year AUD, CUD, stimulant use disorder, other SUDs (including opioid use disorder, sedative/hypnotic/anxiolytic use disorder), more than 1 SUD, and both a psychiatric and SUD. Physical conditions were self-reported lifetime diagnoses of diabetes, heart problems, cancer, other physical problems prior to the MDPS interview. Body mass index (BMI) was also characterized from self-reported height and weight.

Current global assessment of functioning (GAF) was rated by the clinician at the end of the interview¹⁸. Self-reported general health status from the SF-12¹⁹ as well as past year suicidal ideation and attempts were also collected. For suicidal ideation, we included responses to a broad question, “In the past year, have you had any thoughts about taking your own life or just going to sleep and not waking up, or thinking that you would be better off dead?”

Mental health treatment was defined as self-report of ever receiving “professional counseling, medication or other treatment to help with mental health, emotions, or behavior”, as well as “inpatient or residential” or “outpatient treatment for mental health, emotions, or behavior in the past 12 months”. Psychotropic medication prescriptions were defined as being “prescribed by a doctor or health care professional to help with mental health, emotions, behavior, energy, concentration, or ability to cope with stress” during the past year, currently, and type. We defined minimally adequate treatment as taking at least one antimanic agent (a mood stabilizer [i.e., lithium, valproic acid, carbamazepine, lamotrigine or oxcarbazepine] or an antipsychotic), and having at least 4 outpatient visits in the past year consistent with prior research²⁰. Using categories previously defined by Merikangas, et al., 2007, appropriate treatment (i.e., diagnosis concordant), was defined as taking at least one antimanic agent (mood stabilizers or antipsychotics), and inappropriate treatment (i.e., diagnosis discordant) as taking antidepressants or other psychotropic medications in the absence of antimanic agents.¹ SUD treatment was defined as ever receiving “professional counseling, medication or other treatment for alcohol or drug use.” In addition, “inpatient or residential treatment” and “outpatient professional counseling, medication, or other treatment for alcohol or drug use” during the past year was assessed. Certain treatments for SUD were also indicated in the past year (e.g., methadone, buprenorphine).

Sociodemographic characteristics included age, sex at birth, gender identity (female, male, transgender/gender diverse), racial group (American Indian/Alaska Native, Asian, Black or African American, Multiracial, Native Hawaiian/Pacific Islander, White), Hispanic ethnicity, region (Northeast, Midwest, South, West) and urbanicity (urban, rural). Social and economic characteristics included social relationships (marital status, children), highest grade or level in school as well as student status, veteran and active-duty status, criminal justice involvement, employment, family income, housing, and social vulnerability using the social vulnerability metric (SVM, https://svm-tmap.shinyapps.io/SVM-Dashboard-v2/).²¹ Disability and health care coverage were also characterized.

Analytic Plan

Descriptive statistics including frequency distributions and central tendencies with 95% confidence intervals were used to estimate the characteristics of the household sample with BP-I and no MDPS MHD. We calculated adjusted odds ratios (AORs) using logistic regressions with a logit link to compare each of the characteristics between these groups, adjusting for age, sex at birth, race and ethnicity due to differences in prevalence and symptoms by age, sex, racialized and Hispanic ethnicity groups, and also because more women completed the MDPS survey. Mental health disorder comorbidities and treatment were identified using the same descriptive statistics within the sample of individuals with BP-I. Associations between selected covariates and the likelihood of minimally adequate treatment received among individuals with BP-I were also explored. All analyses were conducted using SAS 9.4 survey procedures²² with design weights that accounted for the multistage sampling of the MDPS.

Due to prior findings of heterogeneity of sociodemographic characteristics as well as comorbidities and service use among individuals with bipolar spectrum disorders, we conducted post hoc analyses to evaluate differences by age and sex at birth with Rao-Scott chi-square tests.

The MDPS was approved by the Advarra Institutional Review Board; the present analyses, using anonymized data from the MDPS, were determined not to meet the definition of human subjects research.

Results

Among the MDPS household sample of 4,764 people, 127 met diagnostic criteria for BP-I with a mood episode in the past year (weighted prevalence: 1.5%, 95% Confidence Interval 0.8–2.2) and 2,859 did not meet criteria for any 12-month MDPS MHD (weighted prevalence: 74.8%, 72.3–77.3). Individuals with BP-I were significantly younger than those with no MDPS MHD (Mean=35.3 (SE=3.0) versus 42.8 (0.7) years, F=6.5 (df=91), p=0.01). There were no differences by sex among individuals with BP-I compared to no MDPS MHD. After adjusting for age, sex at birth, and ethnicity, individuals identifying as Asian were less likely than those identifying as White to have a BP-I diagnosis (0.5% versus 4.7%; Adjusted Odds Ratio [AOR]=0.1, 95% Confidence Interval 0.2–0.3) (Table 1).

Table 1.

Demographic characteristics of the household sample with past-year BP-I and no MDPS mental health disorder. ^*N [Weighted % (95% CI)].

	BP-Ia	No MDPS MHDb
N (%) *	127 [1.5 (0.8, 2.2)]	2859 [74.8 (72.3, 77.3)]
Age (M, SE)	35.3 (3.0)	42.8 (0.7)
Age categories
18–25	24 [37.4 (7.4, 67.3)]	291 [14.1 (10.9, 17.3)]
26–44	54 [38.4 (15.4, 61.5)]	1239 [39.8 (35.8, 43.9)]
45–65	49 [24.2 (10.4, 37.9)]	1329 [46.1 (42.0, 50.2)]
Sex at birth
Female	89 [63.8 (42.9, 84.7)]	1636 [48.1 (43.4, 52.8)]
Male	38 [36.2 (15.3, 57.1)]	1223 [51.9 (47.2, 56.6)]
Gender identity
Female	84 [62.6 (41.6, 83.6)]	1624 [48.0 (43.3, 52.6)]
Male	37 [35.3 (14.5, 56.1)]	1206 [51.5 (46.8, 56.1)]
Transgender/Gender Diverse	<10 [1.8 (0.0, 4.1)]	17 [0.3 (0.1, 0.5)]
Racial group
American Indian/Alaska Native	<10 [1.2 (0.0, 2.9)]	34 [1.0 (0.4, 1.6)]
Asian	<10 [0.5 (0.0, 1.1)]	187 [4.7 (3.0, 6.4)]
Black or African American	11 [29.0 (0.0, 59.0)]	296 [13.3 (9.7, 16.9)]
Multi-Racial	<10 [2.3 (0.0, 6.1)]	131 [4.3 (3.2, 5.6)]
Native Hawaiian/Pacific Islander	0 [0.0 (0.0, 0.0)]	<10 [0.9 (0.0, 2.2)]
White	104 [66.2 (37.7, 94.8)]	2128 [72.6 (68.4, 76.8)]
Hispanic ethnicity	16 [8.0 (1.3, 14.7)]	423 [20.4 (14.1, 26.6)]
Urbanicity
Urban	108 [67.6 (36.2, 99.1)]	2447 [83.0 (75.6, 90.4)]
Rural	19 [32.4 (0.9, 63.8)]	412 [17.0 (9.6, 24.4)]

BP-I: Bipolar 1 disorder

MHD: Mental Health Disorder

^*Weighted Percents: To adjust for multistage sampling design

^a.Respondents who met criteria for past year BP-I

^b.Respondents who did not meet criteria for an MDPS MHD (i.e., SSD, MDD, BP-I, GAD, PTSD, OCD, or anorexia nervosa).

Clinical Characteristics (Tables 2–3)

Table 2.

Prevalence of comorbid mental health symptoms, disorders, and treatment in the household sample with BP-I. ^*N [Weighted % (95% CI)].

	BP-Ia
N (%) *	127 [1.5 (0.8, 2.2)]
BP-I Symptoms (past year)
Major Depressive Episode	99 [82.5 (70.7, 94.4)]
Manic Episode	84 [76.6 (64.0, 89.2)]
Major Depressive Episode, only	43 [23.4 (10.8, 36.0)]
Manic Episode, only	28 [17.5 (5.6, 29.3)]
Depressive and Manic Episodes	56 [59.1 (39.3, 79.0)]
One psychotic symptom presentb	51 [22.6 (12.0, 33.3))]
Comorbid Psychiatric Disordersc (past year)
Generalized Anxiety Disorder	50 [53.6 (29.4, 77.8)]
Posttraumatic Stress Disorder	32 [31.3 (10.3, 52.2)]
Obsessive Compulsive Disorder	16 [6.8 (1.8, 11.8)]
≥1 comorbid psychiatric disorder	68 [73.4 (58.3, 88.6)]
Diagnosis of Mental Health Disorder pre-MDPS d	113 [61.2 (33.2, 89.2)]
Bipolar Disorder	96 [62.9 (33.4, 92.3)]
Depression	106 [57.7 (29.2, 86.2)]
SSD	10 [4.9 (0.7, 9.1)]
Other emotional problem	58 [34.9 (17.3 52.5)]
Mental Health Treatment
Ever treatment	119 [75.7 (44.4, 100.0)]
Any Treatment, past year	107 [67.6 (38.7, 96.4)]
Inpatient/Residential, past year	13 [5.4 (0.8, 9.9)]
Outpatient, past year	101 [64.9 (36.9, 92.9)]
Number of visitse	29.9 [14.9 (0.3–59.4)]
Only outpatient, past year	88 [59.5 (33.2, 85.9)]
Psychotropic Prescriptions
Past Year	94 [50.9 (27.0, 74.8)]
Currently	87 [46.0 (22.5, 69.5)]
Any Antipsychotic (AP)	27 [15.5 (4.3, 26.7)]
Any Antidepressant	58 [28.5 (13.5, 43.5)]
Any Mood Stabilizer (MS)	48 [24.8 (10.6, 39.1)]
Lithium	<10 [2.3 (0.0, 4.9)]
Any Anxiolytic/Sedative/Hypnotic	34 [17.7 (5.3, 30.2)]
Any Stimulant/ADHD medication	<10 [5.3 (0.0, 14.8)]
Other	<10 [0.6 (0.0, 1.3)]
Any Psychotropic Polypharmacyf	51 [58.6 (41.4, 75.8)]
Minimally adequate treatmentg	38 [18.7 (7.3, 30.0)]
Appropriate Treatmenth	50 [28.9 (12.5–45.4)]
Inappropriate Treatmenth	33 [13.7 (5.5–21.9)]

BP-I: Bipolar 1 disorder

SSD: Schizophrenia spectrum disorders (schizophrenia or schizoaffective)

^*Weighted Percents: Responses were weighted to adjust for complex multistage sampling design

^a.Respondents who met criteria for past year BP-I

^b.SCID criteria: removed those who only met criteria for catatonic, avolition, diminished emotional expression; removing not better explained by substance use or general medical condition

^c.Respondents met criteria for these mental health conditions in the past year

^d.Respondents self-reported ever diagnosis of these disorders prior to MDPS

^e.Among those with outpatient visits

^f.Taking more than one type of psychotropic (e.g., mood stabilizer with antidepressant) among those currently taking a medication

^g.Current antimanic agent (lithium, valproic acid, carbamazepine, lamotrigine, oxcarbazepine) OR any AP and ≥4 OP visits past year

^h.Appropriate: Mood stabilizers, anticonvulsants, and antipsychotics; Inappropriate: Antidepressants and other psychotropic medications in the absence of antimanic agents

Table 3.

Substance use, treatment, clinical characteristics and functioning in the household sample with BP-I and no MDPS mental health disorder. ^*N [Weighted % (95% CI)].

	BP-Ia	No MDPS MHDb	AORc	AORc (unweighted)
N (%)	127 [1.5 (0.8, 2.2)]	2859 [74.8 (72.3, 77.3)]
SCID Comorbid Substance Use Disorders (SUD) (past year)d,e
Alcohol Use Disorder (ref=no)	19 [33.0 (4.1, 62.0)]	184 [4.4 (3.2, 5.7)]	8.38 (2.88–24.38)	2.57 (1.53–4.31)
Cannabis Use Disorder (ref=no)	15 [8.7 (2.7, 14.7)]	70 [2.0 (1.0, 2.9)]	3.55 (1.25–10.05)	4.98 (2.71–9.18)
Opioid Use Disorder (ref=no)	<10 [0.2 (0.0, 0.7)]	16 [0.4 (0.2, 0.6)]	0.65 (0.07–5.89)	1.38 (0.18–10.68)
Stimulant Use Disorder (ref=no)	<10 [7.8 (0.0, 16.0)]	45 [1.4 (0.7, 2.0)]	8.52 (2.53–28.64)	3.88 (1.68–8.96)
Any SUD (ref=no)	36 [43.4 (15.7, 71.0)]	264 [6.8 (5.1, 8.5)]	8.64 (3.17–23.51)	3.95 (2.60–6.01)
More than 1 SUD (ref=no)	<10 [6.4 (0.0, 14.0)]	47 [1.2 (0.6, 1.7)]	5.65 (1.33–24.06)	2.94 (1.21–7.14)
Substance use (past year)
Alcohol (ref=no)	92 [82.0 (69.6, 94.5)]	2136 [66.1 (61.6, 70.5)]	2.28 (0.90–5.76)	0.82 (0.55–1.24)
Sedative/ hypnotic/ anxiolytic (ref=no)	51 [25.8 (10.4, 41.2)]	349 [6.6 (5.0, 8.1)]	6.33 (2.83–14.16)	5.00 (3.39–7.38)
Cannabis (ref=no)	65 [66.5 (47.9, 85.2)]	756 [18.3 (15.5, 21.1)]	7.23 (3.40–15.36)	2.71 (1.88–3.92)
Stimulants (ref=no)	28 [19.5 (5.8, 33.1)]	255 [4.9 (3.7, 6.1)]	4.19 (1.78–9.82)	2.54 (1.62–3.97)
Opioids (ref=no)	34 [34.1 (12.2, 56.0)]	397 [12.5 (10.4, 14.7)]	3.84 (1.58–9.39)	2.34 (1.54–3.55)
Cigarette (ref=no)	40 [38.8 (15.8, 61.7)]	505 [19.9 (16.7, 23.0)]	2.34 (0.88–6.22)	2.10 (1.42–3.10)
Vaping any substance (ref=no)	52 [33.6 (15.1, 52.1)]	522 [15.4 (13.2, 17.6)]	2.45 (1.04–5.79)	2.85 (1.93–4.20)
Substance Use Disorder Treatment e,f
Ever Treatment	14 [19.3 (0.0, 41.7)]	69 [24.4 (15.1, 33.8)]	2.30 (0.72–7.31)	2.18 (1.01–4.70)
Any Treatment, past year	<10 [11.9 (0.0, 28.9)]	38 [13.3 (6.9, 19.8)]	1.76 (0.35–8.86)	1.31 (0.49–3.49)
Inpatient, past year	<10 [0.5 (0.0, 1.5)]	12 [4.3 (0.0, 9.2)]	0.09 (0.01–1.94)	0.65 (0.08–5.41)
Outpatient, past year	<10 [10.2 (0.0, 25.9)]	32 [11.2 (5.8, 16.6)]	1.82 (0.30–11.22)	0.95 (0.30–2.98)
Suicidal Thoughts and Behaviors (past year)
Suicide Attempt (ref=no)	<10 [2.9 (0.0, 6.4)]	<10 [0.3 (0.0, 0.6)]	7.18 (1.24–41.67)	10.13 (3.22–31.88)
Suicidal Ideation (ref=no)	76 [70.0 (53.0, 87.0)]	330 [6.3 (5.1, 7.6)]	31.38 (14.78–66.60)	11.44 (7.81–16.76)
MDE Onlyg	26 [21.8 (5.9, 37.6)]	--
MDE and Maniag	42 [71.5 (51.9, 91.1)]	--
Mania onlyg	<10 [6.7 (0.0, 14.3)]	--
Current GAF Score (M, SE)	53.2 (1.1)	82.1 (0.5)	F=463.55, p<.0001	F=321.13, p<.0001
With suicidal ideation (M, SE)	51.5 [1.1 (49.4–53.7)]	70.3 [1.3 (67.7–73.0)]	F=78.24, p<.0001	F=79.08, p<.0001
Without suicidal ideation (M, SE)	57.1 [1.6 (53.9–60.3)]	82.9 [0.5 (81.9–83.9)]	F=204.28, p<.0001	F=135.21, p<.0001
With no MDPS comorbidities	56.6 [2.1 (52.5, 60.8)]	83.4 [0.5 (82.4, 84.4)]	F=123.56, p<.0001	F=119.74, p<.0001
With only a SUD comorbidity	55.4 [2.4 (50.7, 60.2)]	64.0 [2.1 (59.8, 68.3)]	F=9.28, p=0.0030	F=7.32, p=0.0073
With only a psychiatric comorbidity	53.9 [2.2 (49.6, 58.3)]	--
With any comorbidity (psychiatric or SUD)	52.4 [1.1 (50.2, 54.7)]	64.0 [2.1 (59.8, 68.3)]	F=21.90, p<.0001	F=46.52, p<.0001
With both comorbidities	50.1 [1.0 (48.1, 52.2)]	--
GAF <50 (significant impairment, ref=no)	37 [18.1 (7.4, 28.7)]	66 [1.8 (0.9, 2.7)]	10.48 (4.36–25.20)	16.85 (10.52–26.98)
GAF Categories
Minimal Impairment (≥71)	11 [4.2 (0.0, 8.8)]	2083 [83.3 (80.9, 85.7)]	0.00 (0.00 0.02)	0.01 (0.00–0.02)
Mild Impairment (61–70)	30 [16.4 (8.1, 24.8)]	492 [10.4 (8.4, 12.4)]	0.11 (0.04–0.30)	0.09 (0.04–0.20)
Moderate Impairment (51–60)	33 [25.6 (9.5, 41.7)]	181 [3.6 (2.6, 4.7)]	0.58 (0.19–1.83)	0.28 (0.13–0.60)
Serious Impairment (41–50)	37 [46.6 (20.8, 72.4)]	79 [2.2 (1.2, 3.2)]	1.49 (0.38–5.94)	0.76 (0.35–1.63)
Pervasive Impairment (≤40) (ref)	16 [7.2 (1.9, 12.5)]	24 [0.4 (0.2, 0.8)]	--	--

BP-I: Bipolar 1 disorder

GAF: Global Assessment of Functioning

^*Weighted Percents: To adjust for complex multistage sampling design

^a.Respondents who met criteria for BP-I

^b.Respondents who did not meet criteria for an MDPS MHD (i.e., SSD, MDD, BP-I, GAD, PTSD, OCD, or anorexia nervosa)

^c.Comparison between BP-I and No MDPS MHD. Adjusted for age, sex at birth, race, and ethnicity.

^d.Respondents met criteria for this substance use disorder in the past year. Includes individuals with 2 + SUDs

^e.Including alcohol

^f.Among those with a past year MDPS SUD

^g.Among those with past year suicidal ideation

Because this study focuses on individuals with past-year BP-I experiencing mood episodes, there were high percentages of individuals with MDEs (82.5%, 70.7–94.4) or manic episodes (76.6%, 64.0–89.2), with a majority (59.1%, 39.3–79.0) experiencing both in the past year (Table 2). Almost a quarter had at least one psychotic symptom in the past year (22.6%, 12.0–33.3). Psychiatric comorbidities were common; 73% (58.3–88.6) were diagnosed with at least one other past-year MDPS psychiatric disorder. Over half had comorbid GAD and one third had comorbid PTSD. Two-thirds of those with BP-I reported that they had received a diagnosis of bipolar disorder prior to the MDPS, while only 21.2% (0.0–50.7) of individuals aged 18–25 had received that prior diagnosis (Supplemental Table 1).

Individuals with BP-I compared to no MDPS MHD had a higher likelihood of using almost all substances in the past year (Table 3). The most used substances were alcohol (82%, 69.6–94.5) and cannabis (66.5%, 47.9–85.2). Compared to people with no MDPS MHD, those with BP-I were more likely to use cannabis (66.5% vs. 18.3%, AOR 7.2, 3.4–15.4), opioids (34.1% vs. 12.5%, AOR 3.8, 1.6–9.4) and sedative/hypnotic/anxiolytics (25.8% vs. 6.6%, AOR 6.3, 2.8–14.2). Individuals with BP-I were over 8 times more likely to have any MDPS SUD than those with no MDPS MHD (43.4% vs. 6.8%, AOR 8.6, 3.2–23.5); they were also 8 times more likely to have AUD (33.0% vs. 4.4%, AOR 8.4, 2.9–24.4).

Over 70% of individuals with BP-I experienced suicidal ideation, over 31 times more likely than those with no MDPS MHD (70.0% vs. 6.3%, AOR 31.4, 14.8–66.6), and almost 3% had a suicide attempt in the past year. Of the less than 10 individuals with BP-I who had a suicide attempt in the past year, they had both a manic and MDE in the past year.

Individuals with BP-I had significantly lower mean GAF scores than those with no MDPS MHD (53.2 vs. 82.1, F=463.55, p<0.01) and were 10 times more likely to have a score less than 50 indicating serious impairment (18.1% vs. 1.8%, AOR 10.5, 4.4–25.2). Although individuals with past year suicidal ideation did have significantly lower GAF scores both among those with BP-I and those with no MDPS MHD, GAF scores were still significantly lower in those with BP-I versus no MDPS MHD. GAF scores were not different between those with and without a mental health or SUD comorbidity among those with BP-I, but were significantly lower for those with no MDPS MHD who had a SUD (83.4 vs. 64.0, F=73.7, p<0.01). Those with BP-I were more likely to report having any physical problem compared to those with no MDPS MHD (33.4% vs. 24.1%, 2.1, 1.0–4.4). Self-reported general health ratings were poor among those with BP-I, and they were over 17.6 times more likely to endorse poor compared to good general health than individuals with no MDPS MHD. (Supplemental Table 2).

Treatment and Service Use (Tables 2–3)

Of individuals with BP-I in the sample, 64.9% (36.9–92.9) had outpatient and 5.4% (0.8–9.9) had inpatient treatment in the past year. When considering treatment by mood presentation, although the majority of individuals with BP-I had both a manic and MDE in the past year (59.1%), only 8% of these individuals had minimally adequate treatment (Supplemental Table 3). Among those aged 18–25 years, only 30.6% (0.0–65.6) received any treatment in the past year (i.e., outpatient, inpatient or psychotropic medication; Supplemental Table 1).

Half (50.9%, 27.0–74.8) of individuals with BP-I received psychotropic prescriptions for mental health in the past year, with 46% (22.5–69.5) currently taking a psychotropic medication. The most common psychotropics were antidepressants (28.5%, 13–5–43.5), mood stabilizers (24.8%, 10.6–39.1), and antipsychotics (15.5%,4.3–26.7). Only 2.3% (0.0–4.9) of individuals with BP-I were taking lithium and only 18.7% (7.3–30.0) received minimally adequate treatment. Over half (58.6%, 41.4–75.8) of individuals with BP-I who were currently taking medication were taking psychotropic polypharmacy (e.g., taking a mood stabilizer with an antipsychotic). Among individuals aged 18–25 years, 18% were currently taking a psychotropic and fewer than 1% received minimally adequate treatment (Supplemental Table 1). Among those with BP-I who also had a SUD, 11.9% (0.0, 28.9) received any SUD treatment in the past year, which was not different than those with no MDPS MHD with a SUD (13.3%, 6.9–19.8, AOR 1.8, 0.3–8.9).

Social and economic characteristics (Table 4)

Table 4.

Social and economic characteristics of the household sample, with BP-I and with no MDPS mental health disorder. ^*N [Weighted % (95% CI)].

	BP-Ia	No MDPS MHDb	AORc	AORc (unweighted)
N (%) *	127 [1.5 (0.8, 2.2)]	2859 [74.8 (72.3, 77.3)]
Highest grade/level of school (ages 21+)
Up to HS or equivalent (GED) (ref=no)	18 [13.0 (3.7, 22.2)]	418 [33.6 (27.1, 40.1)]	0.30 (0.12–0.77)	1.13 (0.67–1.91)
Some college or AA (ref=no)	49 [68.0 (50.0, 85.9)]	733 [31.5 (27.1, 35.8)]	4.24 (2.00–8.98)	1.97 (1.34–2.88)
College Graduate or more (ref=no)	53 [19.1 (6.9, 31.2)]	1604 [34.8 (29.5, 40.2)]	0.43 (0.21–0.89)	0.51 (0.35–0.74)
Student status, current (ages 21+) (ref=no)	12 [25.7 (0.0, 60.0)]	227 [6.7 (4.5, 8.9)]	3.10 (0.78–12.34)	1.02 (0.54–1.94)
Employment (ages 21+)
Employed, past week (ref=no)	67 [36.8 (17.7, 55.9)]	1889 [65.4 (61.0, 69.7)]	0.29 (0.12–0.69)	0.47 (0.32–0.68)
Actively looking for work among unemployed (ref=no)	10 [14.7 (0.0, 29.3)]	167 [16.0 (11.1, 20.9)]	0.30 (0.04–2.37)	0.56 (0.24–1.30)
Family Income
≥$20,000 (ref)	85 [48.2 (24.8, 71.7)]	2392 [79.9 (75.3, 84.6)]	--	--
<$20,000	42 [51.8 (28.3, 75.2)]	381 [15.0 (11.7, 18.3)]	4.86 (1.90–12.43)	3.18 (2.10–4.81)
SVM Score (M, SE)	0.13 (0.22)	−0.26 (0.16)	F=1.45, p=0.2312	F=5.65, p=0.0175
Housing Type
Owned by someone in household (ref=no)	61 [54.3 (29.8, 78.8)]	1842 [66.7 (62.3, 71.2)]	0.78 (0.17–3.55)	0.54 (0.37–0.79)
Rented by someone in household (ref=no)	62 [34.4 (15.0, 53.8)]	975 [31.1 (26.7, 35.5)]	0.84 (0.23–3.05)	1.71 (1.17–2.49)
Other (ref=no)d	<10 [11.3 (0.0, 30.1)]	42 [2.2 (0.7, 3.7)]	6.96 (0.83–58.14)	2.23 (0.77–6.46)
Housing Status
Homeless, past year (ref=no)	<10 [5.9 (0.0, 12.1)]	47 [1.0 (0.7, 1.4)]	5.26 (1.36–20.41)	3.63 (1.64–8.04)
Number of people in household (M, SE)	3.5 (0.4)	3.5 (0.1)	F=0.14, p=0.7050	F=13.52, p=0.0002
Living Alone (ref=no)	41 [12.1 (3.4, 20.8)]	585 [8.0 (6.5, 9.5)]	1.91 (0.84–4.30)	2.12 (1.43–3.14)
Social Relationships
Marital status
Married	35 [34.1 (12.2, 56.1)]	1367 [54.3 (49.8, 58.8)]	0.74 (0.34–1.60)	0.52 (0.33–0.83)
Previously Married	33 [12.2 (3.9, 20.5)]	502 [13.9 (11.2, 16.6)]	0.95 (0.42–2.17)	1.46 (0.87–2.48)
Never Married (ref)	59 [53.7 (29.1, 78.3)]	990 [31.8 (27.7, 36.0)]	--	--
Living with a partner (ref=no)	21 [10.8 (2.6, 19.0)]	338 [10.4 (8.1, 12.8)]	0.64 (0.24–1.70)	0.88 (0.53–1.46)
In a relationship (ref=no)	78 [76.5 (64.7, 88.4)]	1929 [72.1 (68.4, 75.7)]	1.58 (0.66–3.76)	0.76 (0.52–1.10)
Children (ref=no)	64 [68.8 (52.8, 84.9)]	1688 [68.1 (63.9, 72.2)]	1.57 (0.51–4.87)	0.80 (0.54–1.20)
Veteran status (ref=no)	<10 [11.3 (0.0, 30.1)]	181 [6.8 (4.5, 9.0)]	3.06 (0.46–20.16)	1.02 (0.43–2.40)
Currently active duty (ref=no)	0 [0.0 (0.0, 0.0)]	19 [1.4 (0.0, 2.9)]	--	--
Criminal Justice Involvement
Any criminal justice involvement, past yeare (ref=no)	<10 [2.5 (0.1, 5.0)]	71 [2.6 (1.6, 3.6)]	1.04 (0.31–3.54)	2.65 (1.17–6.02)

BP-I: Bipolar 1 disorder

MHD: Mental health disorder

SVM: Social Vulnerability Metric based on individuals’ zip code (positive numbers indicate higher vulnerability) adjusted for population size. (Saulsberry, et al. The social vulnerability metric (SVM) as a new tool for public health. Health Serv Res. 2023 Aug;58(4):873–881)

^*Weighted Percents: To adjust for multistage sampling design

Aged 21+: N=2,879 (BP-I=120, No MDPS MHD=2,759)

^a.Respondents who met criteria for BP-I

^b.Respondents who did not meet criteria for MDPS MHD (ie., schizophrenia spectrum, major depression, PTSD, OCD, generalized anxiety disorder, anorexia nervosa)

^c.Comparing BP-I to No MDPS MHD Detected. Adjusted for age, sex at birth, race and ethnicity. AOR (95% CI)

^d.Other: Owned/Managed by third party such as a college dorm or nursing home, or without payment of rent

^e.Includes arrest, probation, parole, jail stay, and/or prison stay in the past year

To restrict the analyses to participants who were likely to have completed their formal education and to have entered the work force, analyses for likelihood of educational attainment and employment were limited to individuals aged ≥21 years, removing N=107, reducing the sample with BP-I to 120 and no MDPS MHD to 2,759. Those with BP-I were more likely to have completed some college or an AA degree (68.0% vs. 31.5%, AOR 4.2, 2.0–9.0) but compared to those with no MDPS MHD they were less likely to have obtained a college degree or more (19.1% vs. 34.8%, AOR 0.4, 0.2–0.9). Individuals with BP-I were significantly less likely to be currently employed than those with no MDPS MHD (36.8% vs. 65.4%, AOR 0.3, 0.1–0.7).

Among the total sample (including those aged 18–21), those with BP-I were over 4 times more likely to have a family income <$20,000 (51.8% vs. 15.0%, AOR 4.9, 1.9–12.4). They were also 5 times more likely than those with no MDPS MHD to receive disability benefits (30.6% vs. 12.6%, AOR 5.4, 1.8–16.7). They were 4 times more likely to have Medicaid coverage than those with no MDPS MHD (58.8% vs. 19.4%, AOR 4.4, 1.9–10.4). More than 96% of individuals with BP-I had some form of healthcare coverage (Supplemental Table 4).

Likelihood of Minimally adequate treatment (Supplemental Table 5)

Among individuals with BP-I in the sample, exploratory unadjusted analyses revealed that those experiencing both a manic and MDE in the past year as well as individuals with comorbid AUD and opioid use disorder were less likely to receive minimally adequate treatment. Individuals with cannabis use disorder and a prior diagnosis of depression or bipolar disorder were more likely to receive minimally adequate treatment. After adjusting for age, sex at birth, race, and ethnicity (Model I), and further urbanicity, educational attainment, current employment and income (Model II) only those with AUD and cannabis use were less likely to receive minimally adequate treatment, while those with prior diagnoses of depression or bipolar disorders were more likely to receive treatment.

Sex at birth subgroup analyses (Supplemental Table 6)

Clinical Characteristics

Among the sample with past-year BP-I, significantly more males than females reported they had received a lifetime diagnosis of a bipolar disorder (80.3%, 61.1–99.5 versus 52.9%, 17.0–88.8), but there were no differences in prevalence of BP-I symptoms including MDE or mania, or any other self-reported lifetime mental health disorder diagnoses between males and females. Females had significantly higher prevalence of comorbid GAD than males (66.1%, 42.3–90.0 versus 31.5%, 4.4–58.5). The greatest differences by sex were in the prevalence of some SUDs; females had significantly higher prevalence of AUD (46.5% (10.4–82.5) versus 9.3% (0.0–24.1) among males) and any comorbid SUD (56.5% (26.5–86.5) versus 20.2% (0.9–39.5) among males). Women also had higher past-year prevalence of suicidal ideation, over 80% (65.1–95.9) compared to males (51.5%, 20.0–83.0). Although there were no differences between the sexes regarding physical conditions, females had significantly higher mean BMI than males (31.1 (SE: 1.5) versus 28.5 (0.7); F=7.33, p<0.01).

Service Use

More males than females with BP-I in the sample received mental health services with 94.4% (85.6–100.0) of males ever receiving mental health treatment compared to 66.2% (22.1–100.0) of females (F=11.8, p<0.01). More males than females had mental health outpatient visits in the past year (83.4% vs 54.4%; F=4.2, p<0.05). Among people with BP-I and a SUD, significantly more males compared to females received SUD treatment (77.8% versus 7.7% ever receiving treatment, 65.8% versus 1% receiving any treatment in the past year).

Discussion

This study provides a comprehensive and up-to-date description of the clinical, behavioral health service use, social, and economic characteristics of a sample of non-elderly US adults with rigorously diagnosed BP-I disorder who experienced mood episodes in the past year. The study describes current treatment patterns for individuals in the sample with BP-I and updates findings from the NCS-R study, which was fielded two decades ago.

In our sample, individuals with BP-I had lower levels of educational attainment and employment, poorer functioning, and were more likely to be living in poverty than those with no MDPS MHD. These results are virtually unchanged from the NCS-R completed in 2003. Because BP-I typically starts during early adulthood, there is often substantial disruption in all aspects of life including education and employment, and this disruption can persist due to delays in obtaining the correct diagnosis that extends the duration of untreated or inadequately treated illness. This is consistent with our findings that few of the youngest individuals with BP-I in our sample had received a diagnosis of bipolar disorder prior to the study, and had low levels of all forms of treatment, despite higher prevalence of BP-I symptoms.

Mental health comorbidities and substance use among individuals in this sample with BP-I, particularly PTSD and alcohol and cannabis use, were high and consistent with prior studies. The high prevalence of suicidal ideation is serious. The finding that individuals experiencing both a MDE and manic episode in the past year made up all individuals with a suicide attempt suggests that this group may require more intensive clinical attention.

Examination of rates of minimally adequate treatment in our sample is notable in that the rate of receiving minimally adequate treatment was not meaningfully different from the rate found two decades ago in the NCS-R. At the same time, differences in mental health and substance use treatment by sex at birth were striking. More males than females with BP-I received most forms of treatment, despite similar prevalence of most mental health comorbidities, and women had a higher prevalence of substance use disorders and suicidal thoughts and behaviors than men.

Many reasons can help to account for this low treatment rate. Low perceived need and a desire to self-manage symptoms are some of the strongest barriers to seeking treatment²³. Particularly relevant are findings that more women than men report higher stigma-related treatment barriers²⁴. To address this, psychotherapeutic interventions have been designed to support the continued use of psychotropic medications in addition to helping individuals increase quality of life, well-being, and functioning²⁵. At the systems level, psychoeducation and interpersonal and social rhythm therapy²⁶ show the greatest efficacy for BP-I; new uses of dialectic behavioral therapy may also mitigate suicide risk^27,28. Training non-physician clinicians about psychotropics used for BP-I can also enhance team approaches to patient care. Finally, Medicaid and Medicare, which provided over 60% of the health care coverage for individuals with BP-I in our sample, have particularly extensive prior authorization requirements that could decrease access to effective adjunctive psychotropic medications used to treat BP-I^29,30. These treatment gaps suggest persistent high levels of unmet need and underscore the importance of determining and ameliorating barriers to mental health treatment in this population.

Limitations

Only individuals with BP-I who had a mood episode in the past year were detected by the survey because respondents without a past year mood episode were not asked about lifetime mania. Because other epidemiologic surveys also assessed 12-month prevalence of bipolar spectrum disorders, the MDPS is consistent with those prior studies (The World Mental Health Survey Initiative³¹, NCS-R³², NESARC-III², ECA³³). In addition, individuals with schizophrenia spectrum disorders and major depressive disorders were differentiated from those with BP-I in this sample. Because individuals with BP-I are in great need of treatment when experiencing symptoms and due to the variable life course of BP-I with symptoms emerging after years of remission, it is important to understand the circumstances of individuals during symptomatic phases of their disorder. The MDPS had a low overall response rate, which limits the representativeness of our results. This can be attributed to conducting the MDPS during the COVID-19 pandemic; rostering and screening could not be completed in person which leads to higher response rates, as well as the three-stage sampling design. This might have also led to the under-representation of people with BP-I. Although sample weighting was used to adjust for differential non-response based on socio-demographic and geographic factors, these adjustments would not be expected to capture effects of differential non-response associated with psychiatric symptoms other than to the extent that these symptoms are associated with the demographic/geographic adjustment factors. This, in conjunction with evidence from past research that people with psychiatric disorders are less likely than others to respond to community surveys³⁴, suggests that the prevalence estimates found here are likely conservative. However, the low response rate is likely to have a greater effect on the prevalence estimate than on the comorbidity, functional impairment, and treatment profiles of the individuals with BP-I. The adequacy of the treatment measures did not assess individual treatment response, and where or by whom treatments were provided.

Conclusions

This study of individuals with BP-I living in US households revealed that comorbid GAD, PTSD, substance use and comorbid alcohol use disorders were common. Many individuals with BP-I lived in poverty and experienced functional impairment and suicidal thoughts. Few received adequate treatment; women and younger people were least likely to receive any treatment. This portrait has persisted despite a growing number of effective psychosocial and psychopharmacological therapies that have been developed over the last 20 years aimed at mitigating BP-I symptoms and suicide risk. As international treatment guidelines have integrated these new therapies, we had expected to see improvement in the well-being among individuals with BP-I. Further study to determine whether and why these services and newly developed treatments for BP-I are not reaching or meeting the needs of individuals with BP-I in the US is of utmost importance to inform policy makers and providers on addressing these challenges. The under-recognition and under-treatment of BP-I among young adults in this study is concerning given that treatment of BP-I is most effective early in the course of the illness; earlier detection and treatment represent a great opportunity to improve outcomes.

Figure 1.

Past year mental health comorbidities and substance use disorders of those with BP-I and no MDPS MHD.

Figure 2.

Educational attainment and employment among individuals aged 21 and older, and family income of all individuals (aged 18–65 years) with BP-I and No MDPS mental health disorder (MHD)

Highlights.

• Results are from the Mental and Substance Use Disorders Prevalence Study, 2020–2022

• Clinicians interviewed individuals from a sample of US households

• Psychiatric and substance use comorbidities were assessed using the SCID-5

• Mental health and substance use comorbidity was common; rates of treatment were low

• The impact was seen in low GAF scores, employment, and income of those with BP-I