Abstract
Objectives: To examine the long-term outcome of PD patients with psychosis requiring antipsychotic therapy; to explore predictors of mortality, nursing home placement, dementia, and persistent psychosis; and to compare outcomes of those with persistent psychosis vs those whose psychosis resolved.
Methods: Baseline data available from 59 patients enrolled in the PSYCLOPS (PSychosis and CLOzapine in PD Study) trial included age, age at onset of PD, duration of PD and psychosis, character of psychosis, medications, living setting, and scores for Mini-Mental State Examination (MMSE), Unified Parkinson’s Disease Rating Scale, Hoehn and Yahr Scale, and Clinical Global Impression Scale. Longitudinal data were collected 26 months later regarding four outcomes: death, nursing home placement, diagnosis of dementia, and persistence of psychosis. Logistic regression was used to explore whether any baseline characteristics were associated with an increased likelihood of one of these outcomes.
Results: At baseline, 56% of patients had an MMSE score of <25, 12% were in a nursing home, 95% had hallucinations, and 60% had paranoia. On follow-up, 25% were dead, nursing home placement occurred in 42%, psychosis was persistent in 69%, and dementia was diagnosed in 68%. Select baseline characteristics predicted individual outcomes: Nursing home placement was associated with the presence of paranoia and older age; persistent psychosis was associated with younger age at onset of PD and longer disease duration; dementia was associated with older age at PD onset and lower initial MMSE score; no characteristics predicted death. Whether psychosis persisted or not had no significant effect on the development of the other three outcomes. The prevalence of hallucinations at follow-up was not different between groups currently receiving antipsychotics vs those on no treatment.
Conclusions: Psychosis in PD requiring antipsychotic therapy is frequently associated with death, nursing home placement, development and progression of dementia, and persistence of psychosis. Still, it appears the prognosis has improved with atypical antipsychotic therapy based on the finding that 28% of NH patients died within 2 years compared with 100% in a previous study done prior to availability of this treatment.
Psychosis has been a well-recognized disabling complication of treated PD since the early days of levodopa therapy.1,2⇓ It is characterized mainly by the presence of visual hallucinations, which may be people (known or unknown), animals, insects, or inanimate objects but may also be auditory, tactile, gustatory, or cenesthetic (involving viscera).2 The hallucinations may be accompanied by illusions and paranoid delusions, the latter being the most serious of the symptoms, leading to hospitalization and even suicide attempts in some patients.1,2⇓ The prevalence of hallucinations as well as risk factors associated with their occurrence have been the subject of several recent studies. In movement disorder clinics, the prevalence ranges from 21 to 46% depending on the definition utilized.3-8⇓⇓⇓⇓⇓ Even in a community-based study, 25% of PD patients experienced hallucinations.9 Risk factors most consistently associated with the onset of hallucinations include cognitive impairment, older age, longer duration of disease, more severe PD, and depression.4-8⇓⇓⇓⇓ Treatment has improved substantially with atypical antipsychotics,10 and new treatments have been an active area of research.
Little information is available on the long-term outcome of PD patients who develop psychosis. Hallucinations are more frequently the reason for nursing home placement than either dementia or motor impairment in PD patients, and mortality of those patients in nursing homes is high.11,12⇓ But these data were derived from patients who had already been placed in nursing homes, and the studies were performed at a time when atypical antipsychotics were not routinely utilized for this problem. The objectives of this study were to assess long-term outcomes of PD subjects who formerly enrolled in a short-term treatment study of hallucinations and psychosis, to compare outcomes in subjects who no longer hallucinate with those with persistent hallucinations, and to examine possible baseline predictive factors for particular long-term outcomes.
The PSYCLOPS (PSychosis and CLOzapine in PD Study) study was a 4-month trial including a double-blind, placebo-controlled portion and an open-label extension performed by the Parkinson Study Group13,14⇓ in the mid to late 1990s. The study examined the effect of clozapine on psychosis in 60 PD patients. Long-term outcome or mortality data are available on all but one (n = 59). That living cohort is now treated by best medical management according to their treating physicians.
Methods.
Patients.
The PSYCLOPS study design included a 1-month double-blind, placebo-controlled, randomized trial of clozapine therapy in actively psychotic PD patients followed by 3 months of open-label clozapine therapy for all subjects completing the first month. The study examined the effect of clozapine on the psychotic symptoms and motor features of PD. Sixty psychotic PD patients were enrolled by six centers. PD was defined by the presence of three of four cardinal features (rest tremor, muscle rigidity, bradykinesia, and postural instability) and no alternative explanations for the findings. Psychosis was defined as a major psychiatric illness in which reality testing was impaired, typically by the presence of hallucinations and delusions, leading to major communication and social problems.13 This definition excludes patients with visual hallucinations and retained insight, and we also excluded patients so severe as to be unable to tolerate 1 month of placebo therapy. Patients were excluded if they had prior exposure to antipsychotic agents (typical or atypical). Prior to enrollment in the PSYCLOPS trial, PD medications were reduced to the lowest dosage that was deemed tolerable for motor function. All patients had to be on a stable dose of PD medication for 7 days prior to entry. Patients’ medications remained stable during the 4-week double-blind study but could then be adjusted as the investigator deemed necessary. The operational criteria for initial study enrollment were psychosis of at least 4 weeks’ duration and severe enough to require antipsychotic therapy as well as a score of 3 (mild) or greater on the Clinical Global Impression Scale (CGIS) for psychosis severity. Each subject had to have reliable caretakers who could accurately report daily level of function, accompany the subject to each visit, and administer study medication. For other exclusion criteria, see ref. 13. Enrollment in PSYCLOPS was from April 5, 1995, to October 2, 1996 (75% were enrolled by April 9, 1996), with the final patient completing in February 1997.
Data collection.
At the baseline visit (time of enrollment) of the clinical trial, the following data were collected: age, gender, age at onset of PD, duration of PD, duration of psychosis, Mini-Mental State Examination (MMSE) scores, Unified Parkinson’s Disease Rating Scale (UPDRS) total and subscores, Hoehn and Yahr stage (H&Y), CGIS score, psychosis type (hallucinations, paranoia, both), PD drugs and doses currently utilized, and living situation, either home or nursing facility.
On a single occasion in 1999, longitudinal outcome data were collected from each site on each study subject who could be located (n = 59) via standardized data collection forms. The data were collected by the site investigator and coordinator through medical record review and, when possible, patient visits. The investigative sites followed all patients, with one exception, to the final follow-up visit. The date of the last follow-up visit for these patients (between July 1995 and April 1999) was documented. For those who died, the date of death was considered the date of last follow-up. Data on four outcomes were collected: 1) mortality, including date of death and cause; 2) nursing home placement, including reason, if available, and whether they died in the nursing home; 3) dementia, based on clinical grounds and diagnosed by the investigator (from Diagnostic and Statistical Manual of Mental Disorders, 3rd ed.: impairment of long- and short-term memory accompanied by higher cortical dysfunction such as impaired judgment, impaired abstract thinking, aphasia, apraxia, or agnosia that interferes with work, social activities, or personal relationships); 4) persistence or resolution of psychosis (based on information collected by the investigator on the last follow-up visit), including pertinent symptoms (hallucinations or paranoia) and current antipsychotic medication. Other information compiled included the number of hospitalizations and reasons and number of patients on different medications and the doses prescribed. If available, UPDRS, H&Y, and MMSE scores were also provided.
In addition to collating the data above, a retrospective analysis was performed to explore whether any baseline characteristics were associated with an increased likelihood of the four target outcomes. Dichotomized outcome variables (presence or absence) of each outcome were determined at follow-up (mean 22.3 months ± 13.7 for all patients, mean 26.2 ± 12 months for those surviving; range 0.27 to 47 months for all patients, 3 to 47 months for survivors). One additional outcome for this analysis was the occurrence of death or nursing home placement. Logistic regression was used to assess individually the association of potential predictor variables with outcome. The odds ratio (OR) is reported with its 95% CI. Analysis was performed with SYSTAT Statistical Software (SPSS, Chicago, IL). We also compared subjects with persistent hallucinations vs those with no psychosis at last follow-up and examined death rate, nursing home placement rate, number with a final diagnosis of dementia, and number still treated with antipsychotic agents. This analysis was performed using Fisher’s exact test.
All analyses were done on patients with complete data for the variables involved. Those items with substantial missing data at follow-up were not included in any analysis. Data were missing if the available medical records did not disclose in a clear fashion whether the variable was present or absent and no further visits were planned. For instance, if the records did not disclose whether the patient was or was not demented, then the data point was considered missing. Also, if the subject had died and was not seen by the investigator for 6 to 12 months, then the data regarding whether dementia developed or psychosis was still present were unknown and were considered missing. We restricted our analysis to variables that were 90% complete with one exception, dementia data, which were 75% complete. A result of this approach is that slightly different subsets of patients may be involved when looking at different outcome measures. A potential limitation is that certain missing data may be more likely to occur in one of the outcome groups, introducing bias. We intended to minimize this by examining variables with complete or near complete data.
Results.
Baseline and outcome data were collected from 59 of the 60 PD patients enrolled in PSYCLOPS. One patient and family could not be located. Baseline characteristics are shown in table 1. Fifty-two (88%) of the 59 patients were living at home at the time of enrollment, whereas 7 (12%) were in a nursing facility. Of the four primary outcomes, data for mortality and nursing home placement were 100% complete, for persistence of psychosis 92% complete (data missing in 5 subjects), and for presence of dementia 75% complete (data missing in 15 patients). There was some other missing information. Age at onset was available in 55 subjects, PD duration in 54 subjects, and psychosis duration in 45 subjects. Psychosis type was known in 58 patients at baseline and in 30 at follow-up. Final treatment status with regard to antipsychotics was known in 55 of the patients. UPDRS data were available on 58 patients at baseline. Other scale scores and medications were available for all patients at baseline but limited numbers at follow-up, and thus these data were not utilized in the analysis. Data on hospitalizations were available for 42 of the patients.
Table 1 Baseline characteristics of the 59 PD patients evaluated in the study
Fifteen (25%) of the 59 subjects were dead at follow-up. Two of these subjects were from the seven living in a nursing home at baseline, and five of the deaths occurred during the 4-month trial, one other shortly afterward. The mean duration of time from baseline to death was 8.3 months (range 0.27 to 30 months). Of the 15 who died, 9 (60%) were demented, 9 (60%) had persistent psychosis, and 6 (40%) were in a nursing facility. The cause of death was pneumonia in seven (two had other concomitant problems: urosepsis and stroke), myocardial infarction in one, chronic obstructive pulmonary disease in one, and in six patients the cause was unknown. Of the 44 who survived, 30 (68%) were demented, 28 (64%) had persistent psychosis, and 19 (43%) were in a nursing facility. Twenty-five patients (42%) were in a nursing home at follow-up, and none of the original seven who were in that living situation at baseline had been discharged. Reasons were provided for 14 of the 18 patients placed after baseline evaluation and they included the following: caregiver could no longer care for the patient (7), caregiver became ill (2), caregiver died (1), worsening parkinsonism (2), worsening dementia (1), and sleep apnea (1). Of the nursing home patients, seven (28%) died. In the end, 32 patients (57%) had either died or were admitted for nursing home care.
The MMSE score was <25 in 33 subjects (56%) at baseline, which we considered indicative of a diagnosis of dementia. Clozapine did not improve this score in the study.13,14⇓ Of the 44 surviving patients, 30 (68%) were diagnosed with dementia. Eight patients not demented at baseline became demented by the follow-up evaluation. Thirty-seven of the 54 subjects with data (69%) had persistent psychosis, whereas 17 did not. Of the 37 patients who remained psychotic, the type of symptoms was known in 30 of them. Twenty-nine (97%) had hallucinations, eight (27%) were paranoid, and seven (23%) had both.
In follow-up, 40 patients remained on an antipsychotic: 23 on clozapine and 17 switched to other agents. Fifteen were on no antipsychotics, and the treatment status of four subjects was unknown. For those remaining on clozapine, the mean duration of therapy was 13.8 (± 12) months and dose ranged from 12.5 to 100 mg/day. Seven patients required repeat white blood cell counts owing to a decrease (three on more than one occasion), but only two patients withdrew from therapy for this reason.
In the 22 months of follow-up, 23 patients required 42 hospitalizations. The reasons for hospitalizations were pneumonia (nine), orthopedic surgery (six), psychosis (five), worsening PD (four), urinary tract infections (four), heart disease (three), hernia surgery (one), fetal transplant surgery (one), placement of a feeding tube (one), colon cancer (one), syncope (one), sleep apnea (one), PD medication change (one), and unknown (four).
Analysis.
Selected baseline characteristics were associated with individual outcomes. A summary of the significant results with OR and 95% CI is shown in table 2. Significant associations are seen when the CI do not include one. There were no baseline characteristics that were significantly associated with death. Those placed in a nursing home were significantly older and had paranoid symptoms as part of their psychosis. We did an analysis that compared those dead or in a nursing home with those alive and living at home. The former was associated with older age and older age at onset of PD. The presence of dementia at follow-up was associated with an older age and older age at onset of PD plus a lower MMSE score and the presence of an MMSE score that was <25 at baseline. Finally, persistent psychosis was associated with a younger age at onset and a longer duration of PD but not dementia.
Table 2 Summary of significant associations of baseline variables with outcomes from logistic regression analysis
In a comparison of the 37 patients with persistent psychosis and the 17 with no psychosis at follow-up, no significant differences were seen with regard to the endpoints examined. Eighteen of 37 (49%) psychotic patients compared with 7 of 17 (41%) nonpsychotic patients were in nursing homes (Fisher’s exact test, p = 0.77). Nine of 37 (24%) psychotic patients were dead compared with 3 of 17 (18%) nonpsychotic patients (Fisher’s exact test, p = 0.73). Twenty-three of 32 (72%) psychotic patients compared with 7 of 12 (58%) nonpsychotic patients were demented (Fisher’s exact test, p = 0.47). In addition to the five patients with missing psychosis data, there were five psychotic patients and five nonpsychotic patients who had missing dementia data. Finally, 29 of 37 (78%) who remained psychotic and 11 of 17 (65%) nonpsychotic subjects continued to be treated with antipsychotics (Fisher’s exact test, p = 0.50). These findings indicate that drug treatment had no impact on the presence of hallucinations for the long term.
Discussion.
In 1993, a case-control study to examine the most frequent reason for nursing home placement in PD was performed.11 The authors studied 11 patients who had been placed and compared them with 22 who were still living at home. Hallucinations were significantly more common in patients placed in a nursing home, though motor impairment and dementia did not differentiate between the two groups. This study was succeeded by a 2-year follow-up12 to examine outcome, and 100% of the nursing home patients had died, with a mean duration of survival of 15.6 months. They also found that nursing home placement in these patients was permanent. These studies focused on hallucinations in relation to nursing home placement and likely involved the most severely affected cases. In addition, these patients were placed between 1987 and 1991, when therapy with new atypical antipsychotics was not available. The patients in our study were more typical of those seen in practice, living at home, for the most part, requiring antipsychotic therapy and receiving clozapine. They were not the most severe of these cases because we enrolled only those we thought could withstand a month of placebo therapy. In this group of patients, we found a 25% mortality rate, and over approximately 2 years, 42% were placed in nursing homes. In addition, the mortality rate for those in nursing homes was 28 vs 100% in the other study. Certainly, the mortality in this group, though still high, is far better than that previously reported,12 and this difference may relate to living at home and treatment with clozapine and, in some, other antipsychotic agents. The reason for placement primarily related to the inability of the caregiver to continue to provide care. It is well known that hallucinations cause a significant increase in caregiver stress,15 and this may be the reason for admission.
In this study, the patients received clozapine as their initial treatment for psychosis. Clozapine was the first atypical antipsychotic to be approved for use in the United States and at the time of initiation of this study remained the only one available. Since then, four additional agents in that class have been approved. As clozapine has some drawbacks, particularly the 1% incidence of agranulocytosis and the required blood monitoring, each new agent has been tried in an attempt to find an alternative that is less restrictive to use. Despite that, clozapine remains the only atypical antipsychotic proven in double-blind studies to be effective in treating PD-related psychosis without worsening motor features.13,16⇓ Risperidone and olanzapine have both been found, in double-blind studies, to have limited use in PD because of significant worsening of parkinsonism, and in both cases, clozapine was found to be superior.17-20⇓⇓⇓ There are no data thus far on the newest agent, aripiprazole. Finally, although quetiapine has been utilized to treat psychosis in PD and is considered by some to be the drug of first choice in this situation, there are only open-label experiences21-23⇓⇓ and no double-blind trials. Therefore, clozapine remains an important treatment for psychosis in PD, making the results of this study relevant.
There have been other attempts to evaluate the long-term outcome of these patients. One study in 19761 looked at the outcome of 18 PD patients followed for about 6 years, but these patients had a mixture of dementia, agitation, delirium, mania, hallucinations, and other mental symptoms. Five (27%) were placed in nursing homes, five (27%) died, six (33%) were incapacitated but living at home, and only two (11%) were home and semi-independent. Recently, another study24 examined the long-term outcome of 27 PD patients treated with either quetiapine or clozapine for hallucinations. Over the 36-month observation period, 50% were placed in nursing homes. Mortality of patients in nursing homes was 62% compared with 52% in those still living at home. They did not provide the 2-year data. An additional report compared early hallucinations (onset within 3 months) and late hallucinations (onset after 1 year).25 The early hallucinators, made up primarily of patients with diffuse Lewy body disease or AD, had a 5-year mortality rate of 41%, whereas 50% had been placed into a facility. The late hallucinators, made up mainly of PD patients, had 5% mortality and only 7% placement at 5 years. However, the study included only patients with 5-year follow-up so that older, more severely affected, and institutionalized patients were underrepresented.
Persistence of hallucinations after initial occurrence has been noted in several studies.12,26⇓ In a recent study, 89 patients with and without hallucinations were followed prospectively for 4 years to examine what happens to hallucinations with regard to onset and persistence. Only 14% of hallucinators became nonhallucinators, and having hallucinations at baseline was a strong predictor for having hallucinations 4 years later. In addition, antipsychotics did not completely stop the hallucinations.26 This finding indicates the persistent nature of this problem. In our study, despite treatment with clozapine, 69% continued to hallucinate, supporting the results of the earlier study. The persistence (or lack) of psychosis at follow-up did not have an impact on the final outcome of these patients. A similar percentage of psychotic and nonpsychotic patients died, were placed in a nursing facility, were treated with antipsychotics, and became demented. This may suggest that the hallucinations themselves are not necessarily associated with a poor outcome but may be indicators of the development of a high-risk stage of the disease.
As alluded to previously, it is well known that cognitive decline is a risk factor for the onset of hallucinations. Conversely, one study indicated in Discussion that “it is probable that the appearance of hallucinations is a harbinger to cognitive decline,”4 which would imply that hallucinations represent a risk factor for dementia. Our data on dementia have limitations because we used different definitions at baseline (based on MMSE scores) and last follow-up (clinical diagnosis). Admittedly, we also have a considerable amount of missing data on dementia (25%) at follow-up, although the subjects without data were equally spread between the groups with and without persistent hallucinations. In spite of these limitations, we found that patients with hallucinations frequently progress to develop dementia over a 2-year period. This finding has been demonstrated in other studies.24
This study examined risk factors for adverse outcomes in hallucinating PD patients. There were no predictors for death. Older age and the presence of paranoia were risk factors for nursing home placement, whereas older age and age at onset and lower baseline MMSE scores conferred greater risk for developing dementia. On the other hand, younger age at onset and longer duration of disease are risk factors for persistent psychosis. In combination, the data indicated that older patients tend to end up in nursing homes with paranoia or to develop dementia, whereas younger-onset patients tend to continue to have hallucinations and remain in the community. Two reports have indicated that the use of dopamine agonists increases the risk for hallucinations.5,26⇓ This study did not demonstrate any increased risk for a poor outcome in patients actively treated with agonists at baseline, but we did not look at their prior exposure. One of the weaknesses of this study was the small patient number. We cannot adequately exclude some potentially important predictors. A study with larger numbers of patients might uncover other important variables that this analysis missed.
Although it has been demonstrated that atypical antipsychotics provide symptomatic relief for psychotic symptoms, it has remained unclear if they improve the long-term outcome of these patients. However, the prevalence of hallucinations in subjects on antipsychotic treatment was not different from the frequency of hallucinations in untreated subjects at follow-up. If we were to compare our results regarding mortality of nursing home patients with those of the prior study that examined this issue,12 it would seem that intervention does affect outcome. Our cohort had both less mortality and less nursing home placement than the prior study. In addition, because paranoia is associated with a poorer prognosis, our study showed that the percentage with this problem diminished after treatment with clozapine from 60 to 27%, supporting the notion that treatment improves outcome. Given the expense, the chronicity of antipsychotic medication exposure, and the potential risk for aggravated parkinsonism with some agents, the benefits of long-term treatment are important to demonstrate. This study is a first attempt to approach the question, and further longitudinal follow-up of this cohort and others with complete data sets on important outcome measures are needed.
Appendix
The Parkinson Study Group.
Memorial Hospital of Rhode Island, Pawtucket, and Department of Clinical Neurosciences, Brown University School of Medicine, Providence, RI: Joseph H. Friedman, MD; Margaret C. Lannon, RN, MS; Robin A. Davies, BSN. Department of Neurology, Albany Medical College, NY: Stewart A. Factor, DO; Diane L. Brown, RN*; Paul J. Feustel, PhD (statistical analysis). Department of Neurologic Science, Rush–Presbyterian Medical Center, St. Luke’s Hospital, Chicago, IL: Cynthia Comella, MD; Christopher G. Goetz, MD; Kimberly A. Janko, RN. Department of Neurology, University of Rochester, NY: Roger Kurlan, MD; Irene Richard, MD; Irenita Gardner, RN; Nancy Pearson, RN. Department of Psychiatry, Case Western University School of Medicine, Cleveland, OH: Mahmoud Parsa, MD*; Kathleen Large, RN. Department of Neurology, University of Tennessee Health Science Center, Memphis: Ronald Pfeiffer, MD; Sarah Rast, RN. Steering Committee: David Oakes, PhD; Christopher G. Goetz, MD; George W. Paulson, MD; Frederick Marshall, MD; Karl Kieburtz, MD; G. Fred Wooten, MD; Alice Rudolph, PhD. *Deceased.
Acknowledgments
The PSYCLOPS study was funded by the orphan drug division of the Food and Drug Administration under grant no. FD-R-001416-02. The Long-Term Outcome Study was supported by the Albany Medical Center Parkinson’s Research Fund and the Riley Family Chair in Parkinson’s Disease (S.A.F.).
Footnotes
-
Disclosure: S.A. Factor: Novartis unrestricted educational grant and honorarium; J.H. Friedman: Novartis consultant; C.G. Goetz: Novartis unrestricted educational grant; R. Pfeiffer: Novartis consultant and speaker.
- Received December 9, 2002.
- Accepted February 24, 2003.
References
Letters: Rapid online correspondence
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
You May Also be Interested in
Hemiplegic Migraine Associated With PRRT2 Variations A Clinical and Genetic Study
Dr. Robert Shapiro and Dr. Amynah Pradhan