Introduction is present and distinctly establishes the purpose of paper. Introduction is appealing and promptly captures the

Hospital Care for Mental Health and Substance Abuse Conditions in Parkinson’s Disease

Allison. W. Willis, MD, MSCI,1,2,3,4* Dylan P. Thibault, MS,1 Peter N. Schmidt, PhD,5 E. Ray Dorsey, MD, MBA,6 and Daniel Weintraub, MD1,7,8

1Department of Neurology, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA 2 Department of Biostatistics and Epidemiology, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA

3 Leonard Davis Institute of Health Economics, University of Pennsylvania, Philadelphia, Pennsylvania, USA

4 Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania, Philadelphia, Pennsylvania, USA

5 National Parkinson’s Foundation, Miami, Florida, USA

6Department of Neurology, University of Rochester Medical Center, Rochester, New York, USA 7 Department of Psychiatry, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA

8 Parkinson’s Disease and Mental Illness Research, Education and Clinical Centers, Philadelphia Veterans Affairs Medical Center, Philadelphia,

Pennsylvania, USA

A B S T R A C T : O b j e c t i v e : The objective of this study was to examine mental health conditions among hospitalized individuals with Parkinson’s disease in the United States. M e t h o d s : This was a serial cross-sectional study of hospitalizations of individuals aged �60 identified in the Nationwide Inpatient Sample dataset from 2000 to 2010. We identified all hospitalizations with a diagnosis of PD, alcohol abuse, anxiety, bipolar disorder, depres- sion, impulse control disorders, mania, psychosis, sub- stance abuse, and attempted suicide/suicidal ideation. National estimates of each mental health condition were compared between hospitalized individuals with and without PD. Hierarchical logistic regression models determined which inpatient mental health diagnoses were associated with PD, adjusting for demographic, payer, geographic, and hospital characteristics. R e s u l t s : We identified 3,918,703 mental health and sub- stance abuse hospitalizations. Of these, 2.8% (n 5 104, 437) involved a person also diagnosed with PD. The major- ity of mental health and substance abuse patients were white (86.9% of PD vs 83.3% of non-PD). Women were

more common than men in both groups (male:female prevalence ratio, PD: 0.78, 0.78-0.79, non-PD: 0.58, 0.57- 0.58). Depression (adjusted odds ratio 1.32, 1.31-1.34), psychosis (adjusted odds ratio 1.25, 1.15-1.33), bipolar disorder (adjusted odds ratio 2.74, 2.69-2.79), impulse control disorders (adjusted odds ratio 1.51, 1.31-1.75), and mania (adjusted odds ratio 1.43, 1.18-1.74) were more likely among PD patients, alcohol abuse was less likely (adjusted odds ratio 0.26, 0.25-0.27). We found no PD- associated difference in suicide-related care. C o n c l u s i o n s : PD patients have unique patterns of acute care for mental health and substance abuse. Research is needed to guide PD treatment in individuals with pre-existing psychiatric illnesses, determine cross provider reliability of psychiatric diagnoses in PD patients, and inform efforts to improve psychiatric out- comes. VC 2016 International Parkinson and Movement Disorder Society.

K e y W o r d s : Parkinson’s disease; suicide; psychia- try; addiction; epidemiology

Parkinson’s disease (PD) is a common neurodegener- ative disease that is diagnosed in 2% of older adults in the United States.1 PD is associated with psychiatric disorders such as depression, anxiety, and psychosis, which can occur as part of the disease processes or as a treatment side effect.2,3 In addition, PD patients may be at higher risk for impulse-control disorders (ICDs; compulsive gambling, buying, sexual and eating behaviors) and dopamine dysregulation syndrome (compulsive PD medication use),4 and these disorders

———————————————————— *Corresponding author: Dr. Allison Willis, Blockley Hall, 723, 423 Guardian Drive, Philadelphia, PA 19104; [email protected]

Funding agencies: This study was funded primarily by the National Insti- tutes of Health (NIH) and the National Institute of Neurological Disease and Stroke (NINDS) via a Mentored Career Development Award K23NS081087(PI-Willis) and the University of Pennsylvania Perelman School of Medicine Department of Neurology Movement Disorders Division.

Relevant conflicts of interests/financial disclosures: Nothing to report. Received: 4 February 2016; Revised: 19 July 2016; Accepted: 24 August 2016

Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/mds.26832

R E S E A R C H A R T I C L E

1810 Movement Disorders, Vol. 31, No. 12, 2016

have been described as side effects of all classes of dopaminergic medications used to treat the motor symptoms of PD.5

Current epidemiology studies of psychiatric illness in PD primarily contain data from neuropsychiatric research instruments administered to academic center or clinical trial populations. Most studies focus on the prevalence of psychiatric conditions or symptoms and draw conclusions based on populations of 10 to 200, with the largest study to date examining 423 de novo patients enrolled in a specialty center observational study.6,7 This approach provides valuable granular detail on psychiatric symptoms and illness in PD at the expense of offering perspectives on psychiatric ill- ness in female, Asian, Black, or Hispanic individuals with PD, because these groups are underrepresented in specialty center and clinical trial populations. Current data on health care utilization associated with psychi- atric illness have focused on the U.S. veteran popula- tion. A national veteran’s affairs database study found that veterans diagnosed with PD and depression were more likely to have other medical diagnoses (such as stroke, congestive heart failure, diabetes, chronic obstructive pulmonary disease), were more likely to have medical (odds ratio [OR] 5 1.34, 1.25-1.44) and psychiatric hospitalizations (OR 5 2.14, 1.83-2.51), and had more outpatient visits than PD patients with- out a depression diagnosis.8 Two regional veteran’s affairs studies also found increased outpatient health care use and greater comorbid burden among veterans with a recorded mental illness diagnosis.9,10

Although there are numerous studies of mood disor- ders in PD, the data on suicide and PD consist largely of individual case reports of suicidal attempts11-13 and question whether dopaminergic intoxication/with- drawal or deep brain stimulation (DBS) surgery14,15

are contributing factors. Two academic center-based studies separately reported that suicidal ideation or attempts in PD patients correlated with measures of depression and anxiety.16,17 It is unclear whether the increased depression and anxiety experienced by PD patients translates into a greater burden of health care utilization for suicide attempts/ideations.

Current PD care guidelines endorse outpatient screening for symptoms and signs of depression, anxi- ety, ICDs, and psychosis.18 Utilization data on acute care for psychiatric emergencies in the PD population would provide the necessary foundation for evalua- tions of the effectiveness of such guidelines and may generate hypotheses about the community level burden of highly relevant but less-studied mental health condi- tions (eg, substance abuse, suicide attempt).

The aim of this study was to investigate the patterns of acute care for mental health and substance abuse (MHSA) conditions among individuals diagnosed with PD. We used data from the Nationwide Inpatient

Sample (NIS) Database, which contains detailed patient, clinical, hospital, and payer data from hospi- tal discharges from 44 states to characterize and com- pare psychiatric hospitalizations of persons with and without PD. Knowing the patterns of severe psychiat- ric illness in PD would increase public and clinician understanding of disabling aspects this disease as well as provide new potential targets for preventive strategies.

Methods

This study was approved by the institutional review board at the University of Pennsylvania.

Study Dataset

The NIS is the largest all-payer inpatient care health care utilization database in the United States and con- tains data from approximately 8 million hospital stays each year. The hospital universe that the NIS draws from consists of community hospitals, as defined by the American Hospital Association, excluding rehabili- tation hospitals. The American Hospital Association defines a community hospital as a "nonfederal short term general and other specialty hospitals, excluding hospital units of institutions." Veteran hospitals and other federal hospitals are excluded. Hospitals are stratified by census region (Northeast, Midwest, West, South) and location (rural, urban), and then progres- sively by teaching status (teaching, nonteaching), bed size category (small, medium, large), and ownership (public, private not-for-profit, proprietary). A random sample of 20% of hospitals in each stratum is drawn, and all discharges from the sampled hospitals are included in the NIS. Sample weights and statistical programs that account for the stratification and survey design are provided to allow researchers to calculate national estimates and confidence intervals, such that researchers can expect that the NIS data collection process will not impact study results.

Study Population

The study population consisted of adults aged 60 and older who were hospitalized at a NIS community hospital between January 1, 2000, and December 31, 2010. We chose to limit this study to older adults to capture the population most at risk for PD and to minimize the contributions of age-related variability in psychiatric illness, such as the increased risk of schizo- phrenia diagnosis in individuals aged 20 to 29 and dis- tinct psychiatric comorbidity and health utilization patterns in younger PD patients.19 We excluded indi- viduals diagnosed with secondary or drug-induced parkinsonism to limit the impact of coding error. Per- sons with PD were identified using the International Classification of Diseases, Ninth Revision, Clinical

A C U T E P S Y C H I A T R I C C A R E O F P A R K I N S O N D I S E A S E

Movement Disorders, Vol. 31, No. 12, 2016 1811

Modification (ICD-9-CM) codes 332 (PD) or 332.0 (paralysis agitans). Individuals without a PD diagnosis were designated as controls. We extracted patient characteristics (age, sex, race, age, primary insurance/ expected payer) and all recorded inpatient diagnoses. Hospital characteristics available to us included hospi- tal size (defined in the NIS using the number of inpa- tient beds and categorized as small, medium, or large), teaching status, and hospital location (rural/urban).

Our primary event of interest was hospital care for MHSA conditions, identified using the Health Care Utilization Project Mental Health Substance Abuse Clinical Classification Software.20 We identified hospi- talizations with a primary discharge diagnosis of alco- hol abuse, anxiety, bipolar disorder, depression, ICDs, mania, psychosis, psychotic depression, substance abuse, and attempted suicide/suicidal ideation.

Statistical Analyses

The NIS is a 20% stratified probability sample of hospital admissions; therefore, stratification clustering and survey weights are required to calculate national estimates of particular diagnoses. The weighted pro- portions of each mental health/substance abuse condi- tion were compared between inpatients with and without PD. Demographic and hospital characteristic variables were compared by PD diagnosis status using a Pearson chi-square or Mann-Whitney U test. Logis- tic regression models were built to compare the odds of each MHSA diagnosis in PD versus the general inpatient population and examine the associations of individual characteristics (race, age, sex) with a MHSA diagnosis. Our models adjusted for payer (Medicare, Medicaid, private, health maintenance organization, or self-pay), admission type (emergent, elective), and hospital teaching status (teaching, non- teaching). Several states in the NIS withhold race data21; these states were excluded from race/sex analyses.

We performed several sensitivity analyses, consider- ing that coding accuracy and bias can affect claim- based studies. We allowed a MSHA diagnosis in any position (not only the primary diagnosis), performed analyses stratified by hospital teaching status, payer type. We also repeated our analyses in the subset of NIS hospitals, which are also designated as primary stroke centers, which are more likely to have subspeci- alty care. All statistical analyses were performed using SAS version 9.3 software (SAS Institute, Inc., Cary, North Carolina).

Results

Demographic Characteristics

We identified 3,918,703 qualifying hospitalizations for MHSA 2000-2010 NIS data. Of these, 2.8%

(n 5 104,437) involved a person also diagnosed with PD. The majority of MHSA hospitalizations involved white individuals (86.9% in the PD group vs 83.3% in the control group). The PD group contained fewer hospitalizations of black patients (prevalence ratio 5 0.51, 0.50-0.53) and more hospitalizations of Asian patients (prevalence ratio 5 1.32, 1.25-1.39; Table 1). The age distribution of MHSA hospitaliza- tions in the PD group was left skewed: 7% (n 5 9,817) were individuals aged 60 to 64; this pro- portion grew to 41% (n 5 55,014) for PD patients aged 80 years and older. MHSA burden was more evenly distributed across age strata in the general inpatient group (Table 1). These findings agree with previous studies that demonstrate (1) PD prevalence increases with age and (2) psychosis and complicated dementia are more common in older PD patients or in later disease stages.2,18,22 Several studies have reported that women are more likely to have a documented psychiatric diagnosis or use psychiatric care serv- ices.23-25 We found sex differences of MHSA diagno- ses in both inpatient groups, but the magnitude of the difference was less in the PD population (male:female prevalence ratio PD 0.78, 0.78-0.79 vs general 0.58, 0.57-0.58).

MHSA Diagnoses Associated With PD Affective Disorders

As shown in Table 2, affective disorders (anxiety, depression, bipolar disorder, or mania) accounted for 88% of hospitalizations of persons with PD com- pared to 78.8% of psychiatric hospitalizations in the general population (chi square, P < .01). Multivari- able regression models that adjusted for patient, pay- er, and hospital factors found that hospitalized PD patients had greater odds of diagnosis of bipolar dis- order (adjusted odds ratio [AOR] 2.74, 2.69-2.79), mania (AOR 1.43, 1.18-1.74), and depression (AOR 1.32, 1.31-1.34). An admitting diagnosis of anxiety disorder was less likely among PD patients (AOR 0.68, 0.67-0.69).

Addiction

Only 4.2% of PD group hospitalizations were for alcohol intoxication, abuse, or dependence compared with 13.6% in the control group (P < .001). This dif- ference represented a 74% lower likelihood of an alcohol abuse diagnosis among PD inpatients (AOR 0.26, 0.25-0.27; Table 2, Fig. 1). Conversely, the odds of a substance abuse diagnosis were slightly more like- ly in the PD group (AOR 1.06, 1.04-1.09).

Suicide

Not all psychiatric behaviors or diagnoses were more common in PD. Hospitalization for suicide

W I L L I S E T A L

1812 Movement Disorders, Vol. 31, No. 12, 2016

ideation or attempt was less common among PD patients (0.84% vs 0.99% in the general population, chi-square P < .05). However, the adjusted odds of a

suicide-related hospitalization were not statistically different between PD patients and controls (AOR 0.98, 0.92-1.04).

TABLE 1. Population characteristics of older adult mental health and substance abuse hospitalizations, nationwide inpatient sample, 2000-2010

Characteristics (no. of nonmissing data) PD (%) General population (%)

Prevalence ratio (95%CI)

PD vs. general population

Race (n 5 3,918,703)* White 91, 809 (86.9) 3,175,455 (83.3) 1.04 (1.04-1.05) Black 4095 (3.9) 289,341 (7.6) 0.51 (0.50-0.53) Hispanic 5985 (5.7) 223,101 (5.8) 0.93 (0.90-0.95) Asian 1397 (1.3) 38,201 (1.0) 1.32 (1.25-1.39) Native American 304 (0.3) 14,418 (0.4) 0.76 (0.68-0.85) Unknown 1947 (1.8) 72,650 (1.9) 0.97 (0.93-1.01) Age group (n 5 4,998,604) 60-64 9817 (7.4) 960,977 (19.7) 0.37 (0.37-0.38) 65-69 15,511 (11.7) 881,175 (18.1) 0.65 (0.64-0.65) 70-74 21,986 (16.6) 787,493 (16.2) 1.02 (1.01-1.04) 75-79 30,410 (22.9) 799,649 (16.4) 1.39 (1.38-1.41) 801 55,014 (41.4) 1,436,572 (29.5) 1.40 (1.39-1.41) Sex (n 5 4,998,604) Male 58,399 (44.0) 1,787,796 (36.7) 1.20 (1.19-1.20) Female 74,339 (56.0) 3,078,070 (63.3) 0.89 (0.88-0.89) Expected primary payer (n 5 4,991,268) Medicare 118,784 (89.6) 3,838,248 (79.0) 1.13 (1.13-1.14) Medicaid 2804 (2.1) 205,773 (4.2) 0.50 (0.48-0.52) Private insurance 9238 (7.0) 669,272 (13.8) 0.51 (0.50-0.52) Self pay 558 (0.4) 63,791 (1.3) 0.32 (0.30-0.35) No charge 70 (0.05) 7035 (0.1) 0.36 (0.29-0.46) Other 1088 (0.8) 74,607 (1.5) 0.53 (0.50-0.57) Admission type (n 5 4,484,731) Emergent 73,576 (62.4) 2,618,335 (60.0) 0.03 (0.03-0.03) Urgent 23,912 (20.3) 872,784 (20.0) 0.03 (0.03-0.03) Elective 20,006 (17.0) 863,039 (19.8) 0.02 (0.02-0.02) Trauma 161 (0.1) 8006 (0.20) 0.02 (0.02-0.02) Other 169 (0.1) 4744 (0.10) 0.03 (0.03-0.04) Hospital teaching status (n 5 4,976,467) Nonteaching 83,159 (62.9) 2,932,836 (60.5) 1.17 (1.16-1.17) Teaching 49,026 (37.1) 1,911,446 (39.5) 1.05 (1.05-1.06)

CI, confidence interval.

TABLE 2. Mental health and substance abuse hospitalization patterns in Parkinson’s disease versus the general population, Nationwide Inpatient Sample, 2000-2010

Diagnosis

MHSA hospitalizations

PD vs. general population

adjusted ORb (95%CI)

PD General population

n %a n %a

Alcohol abuse 5601 4.2 682,966 14.0 0.26 (0.25-0.27) Anxiety 24,349 18.3 1,232,109 25.3 0.68 (0.67-0.69) Bipolar 14,708 11.1 261,327 5.4 2.74 (2.69-2.79) Depression 77,697 58.5 2,469,428 50.8 1.33 (1.31-1.34) Impulse control 227 0.2 4897 0.1 1.52 (1.31-1.76) Mania 118 0.1 3285 0.1 1.44 (1.18-1.74) Psychosis 996 0.8 29,512 0.6 1.25 (1.17-1.33) Substance abuse 8191 6.2 301,892 6.2 1.06 (1.04-1.09) Suicide 1121 0.8 49,906 1.0 0.98 (0.92-1.04)

CI, confidence interval; MHSA, mental health and substance abuse; OR, odds ratio. aTotal percentage may be greater than 100 because multiple diagnoses are allowed per hospitalization. bAdjusted for sex, age, payer, admission type, and teaching status.

A C U T E P S Y C H I A T R I C C A R E O F P A R K I N S O N D I S E A S E

Movement Disorders, Vol. 31, No. 12, 2016 1813

Impulse-Control Disorders

ICDs are associated with dopaminergic medication use in PD,26 and psychosis in PD may be primary or may occur secondary to PD medication use, infections, metabolic derangements, or acute drug reactions. Hos- pitalization for both ICDs and psychosis were more common in the PD group (ICD: 0.17% in the PD group vs 0.10% in the control group; psychosis: 0.75% in the PD group vs 0.59% in the control group; P < .001 for both comparisons).

Demographic Differences in MHSA Diagnoses

Table 3 displays the results of subgroup analyses that examined the extent to which inpatient diagnoses varied by race and gender between PD and control groups, adjusting for age, hospital, and payer charac- teristics. In general, the associations between PD and mood disorders, bipolar disorder, psychosis, impulse disorders, and alcohol abuse were preserved across race and sex subgroups. However, substance abuse was more likely in white (AOR 1.13, 1.10-1.67) and male (AOR 1.15, 1.11-1.20) PD patients, and less likely in blacks with PD (AOR 0.69, 0.60-0.79). Hos- pitalization with suicide ideation or attempt was increased in the Hispanic PD group (AOR 1.42, 1.06- 1.90). The sensitivity analyses (as described in the Methods section) did not produce adjusted odds ratios that differed substantially in magnitude in direction from our primary analyses (Supplementary Table 1).

Discussion

In this health care utilization study, we examined MHSA hospitalizations of older adults with PD. Psy- chiatric disorders are common in PD and are associat- ed to varying degrees with disease processes (specific neurotransmitters, brain regions, and neural circuits) and PD treatments (dopaminergic medications and DBS). Depression and anxiety may precede PD motor signs by several years,27,28 lending biological plausibil- ity for the high prevalence of these disorders in PD patients.29,30 Psychosis, ICDs, and mania most com- monly occur in the context of treatment with dopami- nergic therapy or other PD treatments (eg, DBS, amantadine, or anticholinergic medications), although psychosis is also associated with the disease process itself. Our data demonstrate that PD patients have dis- tinct acute care needs for mental illnesses, potentially providing new insights about the relationships between PD and psychiatric disorders and informing efforts to improve outcomes in PD.

Measuring the burden of psychiatric illness using hospital discharge data provides a different perspective than psychometric studies of academic center popula- tions. Hospitalization for a given illness not only reflects its baseline prevalence in a population but also the effectiveness of outpatient screening, detection, action, and the relative success of outpatient treat- ment. The higher likelihood of a depression diagnosis in PD patients is consistent with the high prevalence (30%-40%) of depression in PD, but also indicates

FIG. 1. Mental health and substance abuse hospitalizations in PD versus general population. OR, odds ratio. [Color figure can be viewed at wileyon- linelibrary.com]

W I L L I S E T A L

1814 Movement Disorders, Vol. 31, No. 12, 2016

TABLE 3. Mental health and substance abuse hospitalizations in Parkinson’s disease by race and sex, Nationwide Inpatient Sample 2000-2010

Diagnosis

MHSA hospitalizations

PD vs. General Population,

Adjusted ORa (95%CI)

PD General population

No./total no. Prevalence, % No./total no. Prevalence, %

Alcohol abuse White 3808/91,809 0.04 400,051/3,175,455 0.13 0.29 (0.28-0.30) Black 379/4095 0.09 78,065/289,341 0.27 0.27 (0.24-0.31) Hispanic 309/5985 0.05 40,113/223,101 0.18 0.21 (0.18-0.25) Asian 32/1397 0.02 4257/38,201 0.11 0.11 (0.06-0.21) Male 4490/58,399 0.08 511,377/1,787,796 0.29 0.27 (0.26-0.28) Female 1111/74,339 0.01 171,589/3,078,070 0.06 0.31 (0.29-0.33) Anxiety disorder White 16,803/91,809 0.18 823,749/3,175,455 0.26 0.66 (0.65-0.67) Black 557/4095 0.14 52,892/289,341 0.18 0.67 (0.61-0.74) Hispanic 1199/5985 0.20 53,962/223,101 0.24 0.76 (0.70-0.82) Asian 229/1397 0.16 9555/38,201 0.25 0.64 (0.51-0.81) Male 8806/58,399 0.15 349,131/1,787,796 0.20 0.70 (0.68-0.71) Female 15,543/74,339 0.21 882,978/3,078,070 0.29 0.65 (0.64-0.66) Bipolar disorder White 10,601/91,809 0.12 177,984/3,175,455 0.06 2.74 (2.68- 2.8) Black 451/4095 0.11 14,786/289,341 0.05 2.91 (2.62- 3.23) Hispanic 470/5985 0.08 9065/223,101 0.04 2.3 (2.05- 2.59) Asian 111/1397 0.08 1553/38,201 0.04 3.07 (2.29-4.1) Male 6499/58,399 0.11 88,203/1,787,796 0.05 2.9 (2.8-2.99) Female 8209/74,339 0.11 173,124/3,078,070 0.06 2.57 (2.5-2.65) Depression White 53,935/91,809 0.59 1,657,607/3,175,455 0.52 1.28 (1.26-1.3) Black 2291/4095 0.56 114,700/289,341 0.40 1.7 (1.59-1.82) Hispanic 3326/5985 0.56 102,159/223,101 0.46 1.38 (1.3-1.47) Asian 806/1397 0.58 18,476/38,201 0.48 1.66 (1.39-1.99) Male 32844/58,399 0.56 746,793/1,787,796 0.42 1.48 (1.45-1.51) Female 44853/74,339 0.60 1,722,635/3,078,070 0.56 1.15 (1.12-1.17) Psychosis White 698/91,809 0.21 18,993/3,175,455 0.00 1.27 (1.17-1.38) Black 28/4095 0.04 177/289,341 0.00 1.11 (.76-1.63) Hispanic 223/5985 0.22 1319/223,101 0.00 1.35 (0.99-1.85) Asian 13/1397 0.16 229/38,201 0.00 1.63 (0.75-3.51) Male 505/58,399 0.01 11,619/1,787,796 0.01 1.35 (1.21-1.50) Female 491/74,339 0.01 17,893/3,078,070 0.01 1.18 (1.06-1.31) Substance abuse White 5687/91,809 0.06 179,247/3,175,455 0.06 1.13 (1.10-1.17) Black 259/4095 0.06 35,611/289,341 0.12 0.69 (0.60-0.80) Hispanic 275/5985 0.05 13,087/223,101 0.06 0.92 (0.79-1.08) Asian 70/1397 0.05 1925/38,201 0.05 0.73 (0.47-1.14) Male 4384/58,399 0.08 143,014/1,787,796 0.08 1.15 (1.11-1.20) Female 3807/74,339 0.05 158,878/3,078,070 0.05 1.04 (1.00-1.09) Suicide White 793/91,809 0.01 33,467/3,175,455 0.01 0.95 (0.88-1.03) Black 35/4095 0.01 288/289,341 <0.01 1.23 (0.87-1.73) Hispanic 61/5985 0.01 2129/223,101 0.01 1.42 (1.06-1.90) Asian 20/1397 0.01 662/38,201 0.02 0.77 (0.38-1.57) Male 621/58,399 0.01 23,298/1,787,796 0.01 0.97 (0.88-1.07) Female 500/74,339 0.01 26,608/3,078,070 0.01 0.94 (0.85-1.05) Mania White 82/91,809 0.02 2201/3,175,455 <0.01 1.42 (1.23-1.80) Black 0/4095 0.04 165/289,341 <0.01 <0.001 (<0.001->999.99) Hispanic 7/5985 0.02 98/223,101 <0.01 2.50 (0.90-6.87) Asian 0/1397 0.03 35/38,201 <0.01 <0.001 (<0.001->999.99) Male 65/58,399 0.00 1223/1,787,796 <0.01 1.66 (1.27-2.18) Female 53/74,339 0.00 2062/3,078,070 <0.01 1.25 (0.94-1.65) Impulse control White 161/91,809 <0.01 3146/3,175,455 <0.01 1.52 (1.27-1.82)

(Continued)

A C U T E P S Y C H I A T R I C C A R E O F P A R K I N S O N D I S E A S E

Movement Disorders, Vol. 31, No. 12, 2016 1815

that severe depression (requiring hospitalization) is also common in PD. In contrast, a higher risk of bipo- lar disorder associated with PD has not been reported previously. The direction of this finding was robust to several sensitivity analyses, although the magnitude was decreased among hospitals that likely have inpa- tient neurological and psychiatric care available. A greater need for inpatient care for bipolar disorder may reflect the difficulty treating emergent PD in an older adult with a history of bipolar disorder. Basic PD symptom management can precipitate a manic epi- sode in a person with previously controlled bipolar disorder both directly (by the use of dopaminergic medications) or indirectly (by attempting to discontin- ue lithium or neuroleptics). Inexperienced clinicians may misdiagnose PD patients presenting with mania, ICD behaviors, sleep disturbance, hyperkinesis/dyski- nesias, akathisia, or tachyphemia (as a result of the disease process of PD drugs) and a history of depres- sion as having bipolar disorder.

Anxiety disorders are very common in PD, so it might be considered surprising that PD patients were less likely to have an anxiety disorder listed as an admission diagnosis. Depression and anxiety disorders are highly comorbid in PD (up to 80%), but research emphasizes detecting and treating depression and ICDs. Patients with anxiety and depression symptoms may have been coded as having a primary depression diagnosis on admission. Alternatively, anxiety symp- toms occurring in PD may require psychiatric hospital- ization less often.

The prevalence of attempted or completed suicide in PD is not known, although death ideation is com- mon.17 Previous research has not reported higher rates of suicide in PD when compared with the general pop- ulation, even post-DBS surgery.31-33 Our results of rel- atively high burden admission for depression but no increased risk of suicide ideation or behaviors likely reflects practice patterns: PD patients are admitted for depression in the absence of suicidality versus patients

in the general population, where suicidality is more likely to prompt admission.

The association between substance use and alcohol disorders and PD appears complex. PD patients are not widely thought to be at increased risk of substance use disorders; rather, it has been hypothesized that PD patients are risk aversive as a result of disease-related personality changes.34 Substance abuse disorders can be a manifestation of ICDs in PD.35,36 A Swedish reg- istry study that found a history of admission for alco- hol abuse was associated with increased risk (HR 1.38, 1.25-1.53) of a future diagnosis of PD, an intriguing finding potentially linking the biological processes of the 2 diseases.37 The relationship was strongest for the younger onset population, and our previous studies of disabled younger PD patients also found an increased risk of hospitalization for substance abuse,19 together suggesting an interaction between age and substance use disorders in PD patients.

Our subpopulation analyses found that race, and to a lesser extent sex, were moderators of psychiatric admissions when comparing PD patients with the gen- eral population. White PD patients were more likely to be hospitalized for substance abuse, whereas His- panics were more vulnerable to hospitalization for sui- cidal ideation or attempt. These differences may reflect cultural differences in the tendency to seek hos- pital care for mental illness, reduced access of hospi- talization, or be evidence of a greater requirement for inpatient treatment. Another possibility is that our subpopulation data reflects the underservice of PD patients who are black or Asian, groups who experi- ence greater disparities in specialty care38,39 and state- of-the-art care40 compared with whites and Hispanics. Of course, there may be genetic differences in pheno- type, which remain understudied despite the known race and gender-associated differences in risk for developing PD.22,41,42 Future studies are needed to investigate further race disparities in suicide and sub- stance abuse in PD.

TABLE 3. Continued

Diagnosis

MHSA hospitalizations

PD vs. General Population,

Adjusted ORa (95%CI)

PD General population

No./total no. Prevalence, % No./total no. Prevalence, %

Black 12/4095 <0.01 287/289,341 <0.01 2.56 (1.35-4.85) Hispanic 7/5985 <0.01 164/223,101 <0.01 …