Clinical Distinctions Between US Nursing Home Residents With Parkinson Disease With and Without Parkinson Disease Psychosis

Parkinson disease (PD) is a neurodegenerative disease estimated to afflict approximately 1 million Americans and 10 million people worldwide, with about 60,000 new cases diagnosed every year in the United States.^1,2 There are an estimated 5.2% to 6.8% of long-term care (LTC) residents with PD.^3-5 While significant motor features are hallmarks of PD (ie, tremor at rest, bradykinesia, rigidity, and postural instability), the nonmotor symptoms, which can include psychiatric and cognitive disorders, autonomic dysfunction, sleep disorders, and sensory abnormalities, can be more troublesome than motor symptoms in terms of quality of life and financial burden.^1,6-9 Nonmotor symptoms typically precede motor symptoms and the diagnosis of PD by several years.¹⁰ 

PD psychosis is estimated to occur in up to 60% of patients with PD over the course of their disease.¹¹ Symptoms of psychosis include illusions, false sense of presence, formed hallucinations (particularly visual), and delusions.¹² Psychotic symptoms associated with PD are linked to increased dementia, caregiver burden, and mortality, and uncontrolled PD psychosis is a risk factor for premature nursing facility (NF) placement and mortality.^11,13-18 The use of antipsychotic medications to treat the symptoms of PD psychosis has been reported to range from 28% to 50% in a managed care population but has not been evaluated in the NF setting.¹⁹ 

The objectives of this retrospective database analysis were to describe the demographic, functional, and clinical characteristics of US nursing facility residents (NFR) with PD and PD psychosis in a cross-sectional comparison between the groups along with longitudinal comparison of matched cohorts. The aim was to better understand the characteristics of NFR with PD psychosis so as to begin to develop a profile to better inform care.

 This analysis was conducted using linked and de-identified Minimum Data Set version 3.0 (MDS) assessments and prescription claims of US NFR from October 1, 2010, to September 30, 2012, in the Omnicare Inc Senior Health Outcomes data repository of prescription drug claims and MDS data. An exemption from institutional review board full review and a waiver of authorization from the Sterling Institutional Review Board were obtained prior to commencement of this study. 

The MDS is a standardized resident assessment instrument used in the NF setting to develop care delivery and planning, as well as to monitor clinical status, outcomes, and indicators such as falls, fractures, cognition, behavior, rehabilitation potential, resource utilization, and comorbid conditions. It contains scales that effectively summarize assessment domains for resident care, such as the Patient Health Questionnaire-9 and the Brief Interview for Mental Status (BIMS). The MDS has been tested to assess inter-rater reliability for specific items, diagnoses, and quality indicators, as well as for activities of daily living (ADL), Cognitive Performance Scale, and BIMS summary score reliability. 

Cohort Criteria

NFR patients from October 1, 2010, through September 30, 2012, were identified, followed by identifying the NFR with documented PD (PD prevalence) using predefined MDS based criteria ± specific ICD-9 codes. The PD and PD psychosis cohorts were then created using inclusion and exclusion criteria of PD without psychosis and PD psychosis by predefined MDS-based criteria ± specific ICD-9 codes (Table 1). The PD and PD psychosis cohorts were subsequently linked to corresponding prescription claims data for selected treatments of interest. 

Article continues below Table 1.

Procedure

Differences between NFR with PD and PD psychosis were evaluated in two ways. A cross-sectional description of differences between NFR with PD and those with PD psychosis was assessed. Where more than one MDS existed, the most severe rating was used. A subset cross-sectional analysis was performed of NFR with PD alone vs PD psychosis with an active prescription for antipsychotics and/or anxiolytics/sedative hypnotics and having had a fall. In this subset cross-sectional analysis, the MDS with the first reported fall during an active prescription was used; NFR may have had a prescription for more than one medication or medication class simultaneously. Additionally, a longitudinal comparison of a subset of NFR with PD matched by age range and sex to those with PD psychosis was conducted. Differences in clinical characteristics from the first available full MDS assessment to the last available full MDS during the study period (>12 months apart) were examined. 

Statistical Methods

Descriptive measures were used to summarize baseline demographics for each cohort. A chi-square analysis was used to assess the cross-sectional differences between NFR with PD and those with PD psychosis, and differences in falls between the cohorts for those with an active prescription for a psychopharmacologic medication. Longitudinal differences in NFR with PD compared with NFR with PD psychosis were assessed using McNemar test for matched binary data. For comparisons between groups, it was determined a priori that P values <.05 were statistically significant. All statistical comparisons were performed using SAS Version 9.4. 

NFR Patient Flow and Demographics

Figure 1 demonstrates the patient flow in this analysis. Of 300,371 NFR, 15,646 (5.2%) had documented PD. After applying criteria for both cohorts, 11,634 NFR with PD without psychosis and 4012 NFR with PD psychosis were identified. Among the 4012 NFR with PD psychosis, 1723 were excluded for having a diagnosis of Alzheimer dementia and/or schizophrenia, resulting in a PD psychosis cohort of 2289 (14.6%). 

Following linkage of the prescription claims and MDS data, there were 6551 NFR with PD and 2289 NFR with PD psychosis; 6069 (5083 with PD and 986 with PD psychosis) residents did not have matched pharmacy and MDS data. Given the freedom NF and residents have to select their own pharmacy provider in many states, there are instances wherein Omnicare may have had MDS data without prescription claims, explaining the exclusion of 6069 subjects without matching data.²⁰ For the longitudinal analysis, NFR with more than one MDS assessment ≥1 year apart (1522 residents with PD and 491 with PD psychosis with linked prescription claims data), 473/1522 (31.1%) NFR with PD and 473/491 (96.3%) NFR with PD psychosis were frequency matched and evaluated. 

Demographic characteristics of the cohorts in the cross-sectional analysis are depicted in Table 2. NFR with PD psychosis were more likely to be male (52% vs 49%; P=.048) and white (79.2% vs 75.7%; P<.001). The mean age of those with PD psychosis was 78.5 (±9.9) years vs 80.1 (±9.1) years for NFR with PD; P<.001. The majority of both cohorts were admitted from an acute care hospital. However, a higher number of those with PD psychosis were admitted from the community, another NF, a psychiatric hospital, or inpatient rehabilitation facility. The mean body mass index (BMI) (standard deviation [SD]) was approximately 26 (±6.2) in both cohorts; those with PD psychosis were more likely to have unintentional weight loss (P<.001) or weight gain (P=.009) (ie, not on a physician prescribed program to modify weight). Unintentional weight loss or gain was defined as loss or gain of 5% or more in the last month or 10% or more in the last 6 months. 

Article continues below Table 2.

Cross-Sectional Analysis

Some comorbid conditions, recorded as active diagnoses in the last 7 days, included a higher percentage of NFR with PD psychosis having non-Alzheimer dementia, hyperlipidemia, gastroesophageal reflux disease/ulcer, urinary tract infection (within last 30 days), and stroke (Figure 2). NFR with PD psychosis more frequently had additional nonmotor symptoms of PD, including moderate-to-severe cognitive impairment, bowel or urinary incontinence, depression, anxiety disorder, swallowing disorder, behavioral symptoms, and constipation (Figure 3). 

The indicators of pain or possible pain in the last 5 days that were more prevalent in NFR with PD psychosis (48.5% vs 31.7%; P<.001) were nonverbal sounds, vocal complaints of pain, facial expressions, and protective body movements or postures. Of NFR who ranked the intensity of their worst pain over the last 5 days using a descriptive scale (mild, moderate, severe, very severe/horrible), more NFR with PD psychosis reported pain overall (38.4% vs 35.1%, P=.048), despite both cohorts having similar use of prescription analgesics (67.5% with PD psychosis vs 67.6% with PD; P=.912) (Table 3). This may be confounded by the fact that nonverbal sounds as an indicator of pain could also be associated with a hallucination or delusion in a nonverbal resident. 

A greater percentage of NFR with PD psychosis exhibited physical (18.6% vs 5.8%; P<.001) and verbal behavioral symptoms (24.3% vs 7.1%; P<.001). A greater proportion of NFR with PD psychosis had behavioral symptoms that negatively impacted social interactions and participation in activities (6.4% vs 1.6%; P=.029). More NFR with PD psychosis rejected evaluation of care necessary to achieve the residents’ goals for health and well-being (31.2% vs 12.8%; P<.001). More NFR with PD psychosis required extensive assistance/total dependence to perform ADL (78.7% vs 74.4%; P<.001). Additionally, falls after admission were more common in the PD psychosis cohort (57.1% vs 36.9%; P<.001) and resulted in injury twice as often (30.3% vs 15.2%; P<.001). Falls within the PD psychosis cohort were more often characterized as resulting in major injuries (4.8% vs 2.6%; P<.01) (Figure 4). 

Article continues below Table 3.

Table 3 demonstrates that overall use of PD treatments was largely similar between groups. The cohort with PD psychosis received significantly more anticholinergic medications, which can contribute to the development of behavioral and/or psychotic symptoms. NFR with PD psychosis were prescribed more antipsychotics (69.8% vs 22.8%; P<.001), antidepressants (68.7% vs 55.8%; P<.001), and anxiolytics/sedative-hypnotics (55.3% vs 40.7%; P<.001). Use of antipsychotics in the PD without-psychosis cohort may have arisen from the need to manage potentially harmful behavioral symptoms or as adjunctive therapy for depression or bipolar disorder. 

In a subset analysis of NFR with PD psychosis vs PD with an active prescription for an antipsychotic (n=909 vs n=1491, respectively), there was a higher rate of falls in NFR with PD psychosis (37.6% vs 24.5%; P<.001). In a subset analysis of NFR with PD psychosis vs PD alone with an active prescription for an anxiolytic/sedative hypnotic (n=720 vs n=2668, respectively), there was a higher rate of falls in NFR with PD psychosis (27.1% vs 18.7%; P<.001). 

Longitudinal Comparison

For the longitudinal comparison, cohorts were matched by age range and sex, resulting in 473 NFR in each of the PD and PD psychosis cohorts. The demographics of the longitudinal cohort were similar to the cross-sectional cohort, with a mean age of 79 years and a slight majority being female (56.4%). The majority of NFR in each cohort were white (78.9% in the PD psychosis group and 69.6% in the PD group; P<.001). A higher percentage of the NFR with PD psychosis were originally admitted from a psychiatric hospital (3.0% vs 1.1%; P<.001). 

There was a more pronounced weight loss over time in the PD psychosis group than the PD only group (-5.3 lb vs -0.2 lb; P=.004); however, the mean BMI did not change significantly over time between groups. The cohort with PD psychosis had a 1.8% increase in presence of swallowing disorder, while the PD cohort had a 2.4% decrease over time, but there was not a statistically significant difference between groups (P=.70). 

Cognitive function in both cohorts decreased over time, though no significant difference was noted between groups. NFR with PD psychosis had a 3.8% decrease in residents categorized as having no cognitive impairment, a 5.7% decrease in those categorized with moderate impairment, and a 9.5% increase in those categorized as having severe impairment. Likewise, NFR with PD had a 5.2% decrease in residents categorized as having no cognitive impairment, a 1.6% decrease in those categorized with moderate impairment, and a 6.8% increase in those categorized as having severe impairment. Additionally, a higher percentage of NFR with PD psychosis had behavioral symptoms recorded on the MDS at both time points; however, no difference between groups was observed over time. 

A greater percentage of NFR with PD psychosis vs PD only had functional deterioration in ADL (ranging from requiring moderate assistance to extensive assistance or having total dependence on staff) compared with a minimum of 12 months earlier (20.3% vs 11.4%; P<.001). The percentage of residents with falls that caused a major injury increased in residents with PD psychosis, while it decreased in residents with PD only, compared with the first full MDS (11.4% vs 7.1%; P=.042). 

This study provides insight into the characteristics of NFR with PD psychosis and begins to develop a profile to better inform care. Cross-sectional analyses demonstrated that NFR with PD psychosis have more moderate-to-severe cognitive impairment, functional impairment, and behavioral symptoms, as well as falls with major injuries, compared with NFR with PD alone. Longitudinal analyses showed that NFR with PD psychosis had more functional deterioration and falls over time. All of these issues and injuries may lead to an increased burden of care and resource utilization as well as increased risk of poor health outcomes in the NF setting. 

We found an overall prevalence of documented PD among NFR of 5.2%, with nearly 15% of NFR with PD having characteristics consistent with PD psychosis. The rate of PD psychosis seen in this population, while at the lower end of the range, is in the range of reports showing a prevalence of 5% to 75%.^19,21,22 Previous analyses have reported a similar prevalence of PD in NFR, but few have characterized PD psychosis specifically.^3-5,23 This analysis confirms the male prevalence of PD and further reveals a significantly higher male predominance (52%) of PD psychosis in NF.²⁴ This finding is notable given the higher proportion of women in the NF setting.²⁵ 

In this analysis, we compared NFR with PD to those with PD psychosis to understand differences in clinical characteristics, comorbid conditions, cognition, function, and medication use between these two groups. We found that NFR with PD psychosis had more moderate-to-severe cognitive impairment as well as a higher prevalence of severe functional impairment and nonmotor symptoms of PD. 

Fenelon and Alves evaluated the epidemiology of PD psychosis and concluded that, while dopaminergic and anticholinergic medications, older age, and more advanced PD are associated with PD psychosis, a factor associated with the development of hallucinations is dementia.²⁶ This fact is consistent with our findings that nearly 74.4% of the PD psychosis cohort in this study had non-Alzheimer dementia compared with 51.6% of the PD cohort (P<.001). Other studies have documented severe cognitive and functional impairments, older age, and more advanced disease as risk factors for PD psychosis and NF placement.²⁷  

NFR with PD psychosis had more indicators of pain, required extensive assistance or were totally dependent on staff to perform ADL, had more major injuries from falls, more incontinence, and used more antipsychotic prescriptions. Painful symptoms in NFR may result in treatment with psychoactive medications instead of, or in addition to, analgesics, often resulting in adverse consequences. The percent of NFR requiring extensive assistance/total dependence on staff in the longitudinal analysis suggests the greater use of NF resources over time in those with PD psychosis. Together these care areas represent important NF quality measures that facilities are highly motivated to improve because quality measures are posted on the government’s public website, available for payers, patients, and their families to view.²⁸ Poor quality ratings can adversely affect public perception, and thus census, of poorly performing NF. A focus on improvement of care for residents with PD psychosis seems justified. 

Falls remain a significant risk of morbidity and mortality in NFR with PD both prior to and during residence in a NF.^3,29 Further, the risk of falls is greater in NFR receiving antipsychotic medications with PD psychosis. All prescribed antipsychotics included in this analysis now carry a Food and Drug Administration warning about causing somnolence, postural hypotension, and motor and sensory instability that could lead to falls and subsequently fractures or other injuries.³⁰ It is unclear if the higher prevalence of falls in the PD psychosis cohort, 70% of whom received antipsychotic medications, is due to the use of antipsychotics, the psychosis itself, increased cognitive and functional impairments, or perhaps more advanced PD. NFR with PD psychosis were also more often treated with antidepressants and anxiolytics/sedative-hypnotics, which could contribute to the higher prevalence of falls. Falls with resulting injury represent significant changes of condition that may affect the residents’ short- and long-term goals, can lead to hospitalizations, and are an important quality concern for NF. 

A higher prevalence of Centers for Medicare & Medicaid Services (CMS) quality measures (ie, increasing reliance on staff for ADL, injuries from falls, incontinence, and use of antipsychotics) is likely to be associated with higher resource utilization in the NF. The CMS guidance to NF surveyors released in November 28, 2017, indicates that antipsychotics may need to be continued indefinitely for select NFR with enduring, progressive, or terminal conditions, including PD psychosis.³¹ There may be additional burden on residents, NF staff, and caregivers when PD psychosis develops. Treatment of adults with PD psychosis is associated with higher cost across most sites of care (eg, LTC, skilled nursing facility, hospitalization, emergency department, pharmacy, etc.) compared with those with PD alone.^8,9 

Limitations

Retrospective analyses of large databases are inherently dependent on the accuracy and completeness of the recorded data. Under- or overidentification of PD psychosis may have occurred in the absence of an ICD-9 diagnostic code specific to PD psychosis and/or the possibility that NFR developing PD and PD psychosis after admission may not have been formally diagnosed. However, based on the proxy definition of PD psychosis used in this evaluation, clear differences emerged between the PD and PD psychosis cohorts. Given the predefined methods of using the most severe MDS rating for the cross-sectional analysis, if less severe ratings dominated the resident’s NF stay, this could have led to an overstatement of reported parameters. Due to the absence of condition severity parameters in the MDS, we were unable to characterize the severity of PD. While NF admission of adults with PD usually signals severe, and in some cases, end-stage disease, it is possible that a range of disease severities was represented in the NF population studied. We excluded NFR with Alzheimer disease and/or schizophrenia from the PD psychosis cohort, but it is possible that if any miscoding in the MDS database occurred, a few residents with Alzheimer disease and/or schizophrenia may have been in the PD psychosis group. The impact on our results is likely to be minimal.

Lastly, from 2010 to 2012 and going forward, changes in interventions, new drugs, and revision to clinical treatment algorithms occurred that may make extrapolation of these study findings to current practice different from what we report. Evaluations performed using more recent data or newer treatments could result in different findings related to drug treatment and associated risks, including falls with major injuries. Future studies should examine strategies to optimize drug therapy earlier in the course of PD with a goal of reducing the burden of illness and maintaining patients in their communities or returning them home after much shorter NF stays. 

Re-evaluation of strategies to optimize care of adults with PD, prior to the development of psychosis and NF admission, is needed to assist in decreasing the burden associated with this disease. With these goals in mind, clinicians and NF facilities can focus on identifying and optimally managing PD psychosis to improve the quality of care and clinical outcomes for these residents, with the aim of improving CMS quality measures and lowering costs of NF care. 

Authors: Barbara Zarowitz, PharmD, BCPS, BCGP, FCCP, FASCP1 • Terrence O’Shea, PharmD, BCGP2 Andrew Shim, PharmD, JD, MAM3 • Doral Fredericks, PharmD, MBA3 • James C Norton, PhD3

Affiliations:
¹Peter Lamy Center on Drug Therapy and Aging, University of Maryland, College of Pharmacy, West Bloomfield, MI
²Omnicare Inc, Englewood, OH
³ ACADIA Pharmaceuticals Inc, San Diego, CA

Disclosures: 
Dr O’Shea reports research grants from ACADIA Pharmaceuticals Inc. during the conduct of the study and research grants funding from AstraZeneca, Avanir, Eisai, and Sunovion over the last 3 years. Dr Zarowitz reports research grants from ACADIA Pharmaceuticals Inc during the conduct of the study and research grants funding from AstraZeneca, Avanir, Eisai, and Sunovion over the last 3 years. Drs Shim, Fredericks, and Norton are employees of ACADIA Pharmaceuticals Inc.

Acknowledgments:
The authors wish to express their gratitude to Carrie Allen, PharmD, BCGP, BCPS, BCPP, CCHP, who participated in the study design, data collection, and analysis. We would also like to thank Carolyn Atchison, MSc, for her assistance in editing the manuscript.

Address correspondence to:
Barbara J. Zarowitz
4301 Orchard Lake Road, Suite 180-B206
West Bloomfield, MI 48323
Phone: (248) 972-7623
Email: dr.barbarajzarowitz@gmail.com