Early dopaminergic drug-induced hallucinations in parkinsonian patients

Visual hallucinations are images that appear when there is no actual stimulus.¹ In contrast to hallucinations in schizophrenia, in which auditory hallucinations predominate, the major focus of hallucinatory behavior in PD has been on visual experiences.² The primary context in which hallucinations and illusions occur in PD is a late complication of antiparkinsonian medication treatment or occasionally as part of a drug-induced toxic delirium or complication of a superimposed medical illness.³ We describe a series of patients diagnosed initially as having idiopathic PD, but who developed severe hallucinations within 3 months of initiating dopaminergic therapy. Because of the unusual time course for hallucinosis, we compare the hallucinatory syndrome and the 5-year outcome of these patients with a cross-sectional series of patients with chronic drug-induced hallucinations.

Methods. Patients. Of a clinical practice of 787 parkinsonian patients on dopaminergic therapy, we identified all patients who reported hallucinations at a time when the primary diagnosis was PD, had a structured interview focusing on hallucinations at the time of hallucination onset, and had either a 5-year follow-up evaluation that included an updated diagnosis or an autopsy with a final pathologic diagnosis within 5 years of the onset of hallucinations. The diagnosis of PD was based on the presence of at least two of the following signs: resting tremor, cogwheel rigidity, bradykinesia, and postural reflex impairment, at least one of which must be either resting tremor or bradykinesia.⁴ Patients meeting all entry criteria were divided into two groups: those with late-onset hallucinations (LH; beginning after at least 1 year of dopaminergic therapy) and those whose hallucinations were early in onset (EH; within 3 months of dopaminergic treatment).

Data collection. The following were determined by interview: primary type of hallucination (visual, auditory, tactile, olfactory), all types of hallucinations (visual, auditory, tactile, olfactory), content of visual hallucinations (people and animals, objects, lights, other), intensity(vivid with intense colors and definition or vague and poorly differentiated), temporal pattern (primarily nocturnal, both night and day, primarily daytime), and emotional quality (peaceful and pleasant, neutral, or frightening).

Data analysis. Differences between the EH and LH groups were compared using chi-square and two-tailed Student's t-test as appropriate. Significance for all analyses was set at an alpha level of 0.05.

Results. Group characteristics. Seventy patients met all entry criteria: 58 with LH and 12 with EH. The gender ratio was equivalent in the two groups (62% men in the LH group and 67% in the EH group; χ² < 1, p > 0.05), as was the mean age at diagnosis (LH, 59.7 ± 5.3 years; EH, 61.2 ± 9.0 years; t[68] < 1; p > 0.05). Both groups began dopaminergic therapy at approximately the same time after onset of parkinsonism (LH mean, 2.6 ± 1.9 years; EH mean, 2.4 ± 1.7 years; t[68] < 1;p > 0.05). In all patients, hallucinations began during treatment with carbidopa/ levodopa. In the EH group, all patients were on levodopa monotherapy compared with 83% in the LH group (χ² = 2.4, p > 0.05). Among the LH patients, additional dopamine agonists were part of the medication regimen in 18%, anticholinergics in 11%, and amantadine in 5%. No patient received selegiline. The mean duration of dopaminergic therapy before the development of hallucinations in the LH group was 390 ± 99.0 weeks compared with 5.1 ± 4.7 weeks in the EH group (t[68] = 13.4, p < 0.0001). The mean dose of carbidopa/levodopa at the time of hallucination onset was 794 ± 290.5 mg levodopa in the LH group and only 142 ± 87.5 mg in the EH group(t[68] = 7.7, p > 0.0001).

Hallucinations. There were several similarities between the hallucinations in the EH and LH groups. First, all patients in both groups had visual hallucinations, and these were the primary hallucinatory phenomenon in almost all patients (EH, 100% versus LH, 95%; χ² =< 1, p > 0.05). In the few LH patients in whom nonvisual hallucinations (auditory in two patients and tactile in one patient) predominated, all patients experienced at least some visual hallucinations. Second, hallucinations involved images of people or animals (EH, 100%; LH, 91%; χ² 5.4; p > 0.05). Third, hallucinations were vivid in their intensity, color, and definition (EH, 100%; LH, 91%;χ² = 1.4;p > 0.05).

There were several significant differences in the two groups. Hallucinations occurred frequently in the daytime as well as the nighttime in the EH group, whereas this characteristic was atypical of the LH group (69% versus 20%; χ² = 11.3, p > 0.001). Among the LH group, patients hallucinated primarily in the evening hours only, whereas in the EH group hallucinations occurred throughout the day with no predominance in the evening hours. Second, hallucinations were usually frightening, with accompanying suspiciousness and paranoia, in the EH group, but this pattern was rare in the LH group (83% versus 5%; χ² = 40.2, p < 0.0001). Among the LH group, images were usually peaceful and even pleasant. Third, accompanying nonvisual hallucinations were frequent in the EH group and uncommon in LH patients (67% versus 14%;χ² = 29.9, p < 0.0001). In the EH group with additional nonvisual hallucinations, isolated, pure auditory (50%), tactile(50%), and olfactory (38%) hallucinations as well as multiple combinations(38%) occurred. Among the 58 patients in the LH group, only five patients had additional auditory hallucinations, four had added tactile hallucinations, and none had olfactory hallucinations or multiple combinations.

Follow-up evaluations. At the 5-year follow-up, 59% of EH subjects and 95% of LH patients were alive (χ² = 13.1, p < 0.001). The original diagnosis of isolated PD was no longer valid in more EH than LH patients (66% EH versus 14% LH; χ²= 15.8, p < 0.0001). Of the EH patients, all four of the subjects still clinically typical of PD at 5 years were found to have psychiatric diagnoses that preceded the diagnosis of PD by decades. One patient had been diagnosed with bipolar affective disorder at age 22, and one patient had a single psychotic episode at age 32. The two others had been hospitalized for hallucinations before the onset of PD, one 12 years before and one 10 years before the onset of PD. Because of their otherwise typical signs of PD, we considered these patients to have PD and a latent psychosis, with hallucinations acutely unmasked by their dopaminergic drugs for PD. In the remaining eight EH subjects, the diagnosis changed from PD to a different disorder during the follow-up period. The diagnosis was changed to AD with extrapyramidal features in five patients because dementia developed with severe memory loss patterns typical of this diagnosis; three of these patients were confirmed at autopsy. In two other patients the diagnosis of diffuse Lewy body disease was made clinically and confirmed at autopsy in both patients. In the remaining patient, postencephalitic parkinsonism was diagnosed after additional history became available.

In contrast, among the 58 LH patients, the diagnosis of PD was changed in only eight patients during the follow-up period. Based on clinical signs, four were diagnosed with progressive supranuclear palsy, three with striatonigral degeneration, and one with corticobasal ganglionic degeneration. The remaining patients all continued to show signs typical of PD. In the three clinically typical LH patients who died during the follow-up period, PD was confirmed in all patients at autopsy.

Disease course, morbidity, and mortality. In all EH patients, cessation of carbidopa/levodopa led to resolution of the hallucinations, but in two patients with preexisting psychosis, physical injury occurred in the form of hand fracture (patient), shoulder dislocation (family member), and facial trauma (community member) in the midst of their psychotic behavior on dopaminergic drugs. At the 5-year follow-up evaluation, all living patients in both the EH and LH groups were hallucinating actively despite treatment with low doses of neuroleptics (thioridazine 25 to 50 mg/day or clozapine 6.25 to 12.5 mg/day). The PD patients with preexisting psychoses were receiving low doses of dopaminergic medications to control falling, but none of the parkinsonism-plus patients were receiving dopaminergic drugs.

More EH patients than LH patients were admitted to nursing homes during the 5-year follow-up period (EH 50% versus LH 7%; χ² = 14.5, p < 0.0001). In the EH group, two patients with prior psychosis and four of the parkinsonism-plus patients (two AD patients and two patients with diffuse Lewy body disease) required sheltered care, whereas among the LH group only four patients were placed in nursing homes. All EH patients who died were living in a nursing home at the time of death.

Discussion. Visual hallucinations that occur early in the course of parkinsonism have not been reported extensively and have never been studied systematically. Early reports, primarily from the French literature of the nineteenth century, alluded to occasional hallucinatory behavior in untreated PD, but these accounts are few.^5,6 We found that patients with early hallucinations fell into two distinct categories: those with a prior psychosis, in each case forgotten or concealed from the neurologist, and those with another parkinsonian syndrome, usually diffuse Lewy body disease or AD. Because accurate premorbid diagnosis of PD remains problematic without a biological marker, clinical criteria that support or reject accurately the diagnosis of PD are particularly valuable for treatment and research. Most of our patients with LH retained their clinical diagnosis of PD, whereas we revised or amended the diagnosis in all those patients with EH. Our data, which demonstrate that early hallucinations occurred in a variety of non-PD conditions and in PD that accompanies another underlying illness (psychosis), support this observation and emphasize that early hallucinations should steer the diagnostician specifically away from the diagnosis of PD as the sole condition.

The two most common alternate diagnoses were diffuse Lewy body disease and AD. In diffuse Lewy body disease, parkinsonian features of bradykinesia and rigidity are typical, but early dementia and hallucinations usually occur before motor impairment becomes marked.⁷ In the clinical-pathologic report by McKeith et al.,⁸ 80% of these patients had visual hallucinations, 45% had auditory hallucinations, and 80% were delusional. The quality and content of these hallucinations has not been delineated further. In a study using autopsy-based diagnostic criteria, Litvan et al.⁹ found that early hallucinations were among the three clinical features that most accurately differentiated diffuse Lewy body disease and PD.⁹ Both our patients fit their criteria for diffuse Lewy body disease after 3 years of follow-up, with the diagnosis confirmed at autopsy. Likewise, patients with AD and other degenerative disorders hallucinate and have extrapyramidal findings that resemble PD.¹⁰

The other major group of patients with EH had typical PD, but also had an underlying latent psychotic illness that was unmasked by the introduction of levodopa. A detailed description of this "unmasking" syndrome has not been published previously, likely reflecting at least in part its infrequent occurrence. If considered independently, the prevalence rates of psychosis are 1,300:100,000 and of PD are 150:100,000, resulting in an estimated joint probability in 0.2% of PD patients.^11,12 The higher frequency in our population may reflect the selection factors and biases discussed later. Even if uncommon, however, the fact that the psychotic behavior was dangerous to patients, their family, and community members underscores the importance of its rapid recognition and assertive intervention. The emergence of novel antipsychotic medications like clozapine offer a potential treatment along with dopaminergic drugs to manage psychoses without aggravation of PD motor features.

Because our patients were identified from a specialty referral practice, and because patients were selected for a 5-year duration of follow-up, our observations may have been subject to several potential biases and may not apply directly to all persons with parkinsonism. First, we cannot comment on the prevalence of either of these two hallucinatory syndromes in PD. Because our focus was on the comparative characteristics of the hallucinations and the long-term follow-up, we specifically selected only those patients with complete data. Patients who were not available for evaluation at 5 years were not studied. Thus, it is likely that older persons, those with more severe disease, and those requiring institutionalization for any reason are underrepresented in our study. Because we selected subjects retrospectively using complete hallucination interviews, EH patients may be overrepresented in this series because the treating clinician might have been particularly motivated to obtain data on patients with an unusual temporal pattern of hallucinations.

We found that several clinical features made the patients with early hallucinations particularly distinct from those with the classic late-onset, drug-induced hallucinations. The frequent daytime presence of hallucinations, the usual combination of visual with other hallucinations (tactile, olfactory, and auditory), and the predominance of a frightening quality with paranoia were all features that are not reported regularly in the literature or found in our series of patients with hallucinations after several months or years of dopaminergic therapy. The early development of hallucinations, often occurring within days of the initiation of dopaminergic therapy, suggested a presensitization to dopaminergic stimulation. Patients with PD can become acutely intoxicated by dopaminergic drugs given in high doses, but in all cases these patients were exposed to very low carbidopa/levodopa doses. We suggest that the biochemical substrate of underlying psychotic illness or the diffuse pathologic involvement of the parkinsonism-plus syndromes caused these patients to be particularly vulnerable to acute behavioral decompensation.

The primary focus of our study was not the LH patients, but several other cross-sectional studies have reported prevalence rates of hallucinations in PD ranging from 5 to 50%, depending on the group assessed and the methods of ascertainment.^13-15 Although hallucinations are often considered a levodopa effect, they occur with all dopaminergic drugs, including agonists, deprenyl, and antidepressants with dopaminergic activity.^13-15 Prevalence rates increase with the years of chronic drug exposure,¹⁶ and dementia has been considered a risk factor.¹⁷ Tanner et al.¹⁶ found that exposure to multiple drugs and older age were significantly more likely to occur in patients with hallucinations than without hallucinations. PD duration, medication treatment duration, motor severity, and depression have not been linked consistently to hallucinations. In our series both the EH and LH patients persisted in having visual hallucinations at the 5-year follow-up, documenting the chronicity of this phenomenon.

We have studied previously¹⁸ the clinical impact of hallucinations in the context of risk factors for nursing home placement in patients with advanced PD. Using a case-control methodology, we examined patients who entered nursing homes and compared them each with two PD subjects remaining in the community.¹⁸ Each patient was matched with a control subject for age, gender, and PD duration. We compared motor disability, dementia, and hallucinations as putative risk factors for nursing home placement. Both groups were severely disabled with regard to their motor skills and both had moderate cognitive impairment. The only differentiating features of statistical significance, however, were the presence and the severity of hallucinations in the patients. Eighty-two percent of the patients hallucinated regularly compared with only 5% of the PD community-dwelling control subjects. Whereas that study included only patients with LH, our follow-up of the EH group at 5 years also revealed a particularly high admission rate (50%) to nursing homes or chronic sheltered care facilities. Furthermore, all EH patients who died in the follow-up period died in a nursing home. These data reinforce prior conclusions that hallucinating patients are poorly managed at home and are institutionalized frequently. The short life expectancy of hallucinating parkinsonian patients placed in nursing homes makes the identification and control of hallucinations a priority in PD management.¹⁸ The occurrence of early hallucinations in parkinsonian patients receiving dopaminergic therapy can alert clinicians to alternative diagnoses that may improve clinical management and estimates of prognosis.