The Malignant Course of “Benign Hallucinations” in Parkinson Disease

Objective  To monitor progression of “benign hallucinations” in Parkinson disease (PD).

Methods  We searched our data repository for subjects with PD with 3 sets of neuropsychological testing during 3 years and Unified Parkinson's Disease Rating Scale thought disorder scores taken at 4- to 12-month intervals during this period. We found 48 patients with benign hallucinations, defined as a thought disorder score of 2 (benign hallucinations, insight retained), receiving no treatment for hallucinations. We followed up thought disorder scores under best medical management for at least 3 years or until a thought disorder score of 3 (loss of insight) or 4 (delusions) occurred. In subjects whose thought disorder scores remained at 2, we assessed neuroleptic use and decreases in PD medications to abate hallucinations.

Results  Most subjects (81%) progressed to thought disorder scores of 3 or 4. In 7 (78%) of 9 subjects who retained a thought disorder score of 2, PD medications were reduced to treat hallucinations, and 3 subjects (33%) also received neuroleptics. If the composite end point (any of the criteria) was used, 96% of subjects progressed, with only 2 subjects continuing with stable, untreated benign hallucinations.

Conclusions  Because hallucinations progress, the concept of benign hallucinations is prognostically misleading. Though hallucinations with retained insight may be “benign” for the moment, they portend serious consequences. The term benign hallucinations of PD should be considered generally unsound and dropped from operative vocabulary.

Hallucinations are a frequent complication of Parkinson disease (PD) and occur in approximately one third of patients with chronic disease undergoing dopaminergic therapy.¹ Longitudinal studies demonstrate that they are generally chronic, progressive, and associated with nursing-home placement.^2,3 On the other hand, the term benign hallucinations is a commonly used phrase in reference to PD, connoting hallucinations of inconsequential import. The Unified Parkinson's Disease Rating Scale (UPDRS) uses this term in item 2 “thought disorder,” equating the term with retained insight.⁴ The phrase “benign hallucinations” is inherently ambiguous, because benign can refer to the current status of a problem or to its prognosis. To examine the question of whether benign hallucinations have a benign prognosis, we studied a series of patients with detailed neuropsychological testing and UPDRS thought disorder scores of 2 (benign hallucinations) and followed up the patients longitudinally to assess changes in hallucinations under best medical management.

The Rush Movement Disorder Parkinson’s Disease Registry includes outpatient UPDRS data on all patients with PD at all visits, generally spaced 4 to 6 months apart. At all visits, the UPDRS motor examination score is obtained as part of standard care. In addition, patients are routinely scheduled for annual neuropsychological testing to coincide with a clinic visit, and data on cognitive status, dementia, hallucinations, and depression are available to health care professionals when they see the patient on that visit. Dementia was systematically defined according to National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer's Disease and Related Disorders Associations criteria with patients having a history of cognitive decline and evidence of impairment in memory and at least 1 other cognitive domain.⁵

We excluded patients diagnosed with other parkinsonian syndromes at any visit. To decrease the likelihood of including subjects with dementia with Lewy bodies, we excluded patients with hallucinations within 3 years of diagnosis or less than 1 year of dopaminergic treatment before hallucinations began. To examine long-term stability or change in subjects with benign hallucinations in the context of best medical practice, we only considered patients with at least 3 sets of annual neuropsychological tests (baseline, 1 year, and 2 year) and a baseline UPDRS thought disorder score of 2 who were receiving no neuroleptics for control of hallucinations. Our interest in having longitudinal assessments of cognitive status was based on prior reports that dementia is the most prominent behavior associated with hallucinations and, therefore, loss of insight (thought disorder score of 3) could relate closely to concomitant global cognitive decline.^1,6

The primary outcome event was progression of the thought disorder score from a baseline score of 2 to a score of either 3 (hallucinations with loss of insight) or 4 (delusions). We followed up patients for at least 2 years or until this target event occurred. In subjects with thought disorder scores remaining at or lower than 2 during the observation period, we also assessed whether and when neuroleptics were prescribed for treating hallucinations or if PD medications were reduced because of hallucinations.

Data were summarized by means and standard deviations or counts and percentages. Kaplan-Meier survival curves calculated median times for 4 definitions of lost benignity: for the entire sample, time to thought disorder score progressing to 3 or 4; for subjects whose thought disorder score remained at 2 throughout the study, time to the health care professional's decision to decrease PD medicine to abate hallucinations and time to the prescribing of neuroleptics; and for the entire group, a composite end point of the first of the 3 events.

Forty-eight patients, 35 men and 13 women, fulfilled the inclusion criteria. At study entry (first office visit with a thought disorder score of 2), patients' mean (SD) age was 69.0 (6.4) years (range, 55-79 years), and their mean (SD) PD duration was 12.0 (6.2) years (range, 2-28 years). The mean (SD) UPDRS Part III (motor) score was 39.7 (9.02), and all received carbidopa/levodopa (mean [SD] daily levodopa dose, 677 [189.3] mg). Sixty-three percent received agonists (mean [SD] pergolide mesylate equivalent, 1.74 [0.84] mg/d). Other agents were infrequently prescribed (anticholinergics, 0%; amantadine hydrochloride, 6%; selegiline hydrochloride, 27%; catechol-O-methyl-transferase inhibitors, 0%). No patient had brain surgery prior to or during the study, and none had a prior or existing psychiatric comorbid condition associated with hallucinations. Cognitively, at study entry 15% had dementia.

During the 3-year follow-up, 39 (81%) of 48 progressed to thought disorder scores of 3 or 4. In 7 (78%) of 9 of those whose thought disorder score remained at 2, the treating movement disorder neurologists reduced PD medications because of hallucinations, and for 3 (33%) of the 9, they prescribed neuroleptics to treat hallucinations. If the composite end point (any of the criteria) was used, 95% progressed beyond benignity, with only 2 subjects continuing with stable, untreated benign hallucinations.

We compared the motor UPDRS scores when the thought disorder score changed to 3 or 4 vs the score at 3 years for those whose thought disorder scores remained at 2. Scores were slightly, but not significantly, higher than at study entry in both groups (those whose thought disorder scores increased, mean [SD] end UPDRS score, 39.92 [9.48] vs baseline 39.59 [8.96] [P = .48 (paired t test)] and those whose thought disorder scores remained at 2, mean [SD] end UPDRS score, 41.11 [10.40] vs baseline 40.22 [9.82] [P = .17 (paired t test)]). The 2 groups did not differ significantly at baseline (P = .85, t test) or at end point (P = .74, t test). At end point, the mean levodopa dose, agonist use, mean agonist dose, and use of other drugs did not differ significantly from baseline (all P>.50) in either group.

We examined 2 questions related to the association between dementia and the evolution of thought disorder scores to 3 or 4: whether subjects with baseline dementia (n = 7) carried a higher risk for developing malignant hallucinations than the rest of the group (n = 41); and whether subjects who developed thought disorder scores of 3 or 4 (n = 39) had a higher prevalence of dementia at that time compared with those whose thought disorder scores remained at 2 (n = 9). The overall frequency rate of dementia increased during the study (15% at baseline vs 33% at the end of study; P = .003 by the McNemar test). Those with baseline dementia were not more likely to develop thought disorder scores of 3 or 4 than subjects without baseline dementia (86% vs 81%; P>.99 by Fisher exact test). Thirty-eight percent of subjects who declined to thought disorder scores of 3 or 4 had dementia at their end point visit compared with an 11% prevalence rate of dementia among those whose thought disorder score remained at 2 at their 3-year visit. The higher the rate of dementia among those with higher thought disorder scores, however, was not statistically significant, so there was no specific association between malignant hallucinations and concurrent dementia (P = .24 by Fisher exact test). One of the 9 subjects whose thought disorder score remained at 2 but who required neuroleptics or PD medication reduction had dementia both at baseline and at end point. Neither of the 2 whose thought disorder scores remained at 2 at 3 years and who required neither neuroleptics nor PD medication reduction had dementia at either point.

The time from first office visit with a recorded thought disorder score of 2 to the primary target end point of conversion to thought disorder scores of 3 or 4 was short overall (median time by Kaplan-Meier curve, 16 months) (Figure). For the 9 subjects with thought disorder scores remaining at 2, the median time to lowering of PD medications was 22 months. The 2 patients who never reached a target end point and continued to have benign hallucinations without medication changes to treat hallucinations were followed up for 41 and 45 months.

Hallucinations are problematic to study because the health care professional must elicit their description from the patient. Most cross-sectional studies of hallucinations report their occurrence in approximately one third of subjects with PD receiving long-term dopaminergic therapy.¹ These estimates are likely to be underestimates, however, because patients may not volunteer such information easily when the phenomenon itself is strange, disconcerting, and “crazy.” The archetype of fully developed images of people or animals is less frequently encountered than fleeting images in the lateral visual fields (de passage) or very vague shadows.⁶ Most health care professionals concentrate on visual hallucinations, which are the most common of these phenomena, but tactile, auditory, and olfactory hallucinations also occur, and their documentation can be missed if not specifically sought.⁷ Our regular office practice includes a UPDRS thought disorder rating at each visit so that patients are used to these questions, though reticence to endorse hallucinations may still exist.

We view the distinction between hallucinations with or without retained insight as an important operative division, though our data argue against the utility of “benign” as a designation for the former. In the context of full recognition that a sensory phenomenon is false and only related to medication or disease, hallucinations can be of little concern to patients.⁸ In this sense, we concur with the concept of benign as a synonym for “mild.” The cancer and general medical literature, however, have introduced the concept of a safe longitudinal outcome with the term benign, and this concept for hallucinations in PD is clearly contrary to our data. Although hallucinations with retained insight frequently occur in patients with PD, the likelihood that the hallucinations will remain this mild is low. For this reason, we consider the term misleading. Benign hallucinations has been dropped from the working draft of the UPDRS revision, currently in clinimetric testing.⁹

To remove single-center practice patterns from our primary analysis, we chose the strict change in thought disorder score as our outcome of focus. Our 2 secondary target events, however, are practical measures of the loss of benignity because the health care professional's decision to reduce PD medications because of hallucinations or to add a neuroleptic drug are both interventions that can potentially cause further motor decline. We admit that our primary investigator (C.G.G.) is particularly vigilant about hallucinations and their treatment, so his threshold to treat them with pharmacological changes was likely low. The patient population, however, was derived from 6 movement disorder neurologists, and his patients composed only 15% of the sample.

We acknowledge that our study entry requirement of 3 neuropsychological examinations during 3 years may have oversampled patients with more cognitive and behavioral problems, so that they and their neurologists were more vested in ensuring compliance with our standard policy of administering neuropsychological testing on an annual basis. We currently do not have a tracking system to check if patients declined their annual recommendation to be tested, so we cannot estimate the impact of this restriction. We were interested, however, primarily to test associations of concurrent dementia at the time of conversion from benign hallucinations to malignant hallucinations and therefore considered longitudinal cognitive examinations pivotal to our goal. For the simpler assessment of baseline dementia as a risk factor for evolution of hallucinations from benign to malignant, our population would have been much larger, and we are currently examining this issue in this broader sample.

Dementia is a frequent comorbidity with hallucinations,^1,6,8 but we did not find that baseline dementia increased the likelihood that hallucinations with insight would evolve into hallucinations with loss of insight. Further, the prevalence of dementia over time was not specifically associated with the development of hallucinations with loss of insight. These data corroborate earlier studies on the association between cognitive deficits and hallucinations but suggest that dementia is not a primary determinant of the severity of hallucinations.

Using our 3 definitions of loss of benignity, only 2 subjects remained in the benign hallucination category by 3 years. This observation underscores the serious implications of hallucinations over time and their impact on the management of chronic PD. We conclude that benign hallucinations can stay benign, but this stability is atypical in a tertiary care practice. We cannot comment on patients from a community-based or general practice–based sample, but our center services a large metropolitan area with no access restrictions by insurance or public aid.

The time to a change from benign to malignant hallucinations was variable, and although the median time was less than 2 years, some patients converted more than 3 years after study entry and some converted within 2 months of observation. We emphasize that our data cannot be used to estimate the total duration of benign hallucinations because we did not have consistent or similarly retrieved data on most patients prior to enrollment. Therefore, they may have had benign hallucinations for long periods before we started our assessments. Nonetheless, for the practicing health care professional who can meet a patient at any time in the course of PD, the study indicates that patients who have benign hallucinations and are not being treated are likely to progress in severity. We cannot comment on differential progression patterns in patients who are treated with neuroleptics when their thought disorder scores are 2 because neuroleptic treatment was a baseline exclusion criterion.

Our findings reinforce our larger and longer study covering 6 years and enrolling patients with PD spanning the full gamut of thought disorder scores at baseline.¹⁰ In that study, hallucination prevalence and severity increased over time. The methods of that study used a special hallucination inventory. In spite of the different methods and patient population, the similarity of overall findings prompt us to consider the concept of benign hallucinations a misnomer that should be dropped from the operative PD vocabulary.

Correspondence: Christopher G. Goetz, MD, 1725 W Harrison St, Chicago, IL 60612 (cgoetz@rush.edu).

Accepted for Publication: January 4, 2006.

Author Contributions:Study concept and design: Goetz, Leurgans, and Stebbins. Acquisition of data: Goetz, Bernard, and Stebbins. Analysis and interpretation of data: Goetz, Fan, Leurgans, and Stebbins. Drafting of the manuscript: Goetz. Critical revision of the manuscript for important intellectual content: Goetz, Fan, Leurgans, Bernard, and Stebbins. Statistical analysis: Fan, Leurgans, and Stebbins. Obtained funding: Goetz. Administrative, technical, and material support: Bernard. Study supervision: Goetz and Leurgans.