Literature review

A systematic literature search using MEDLINE and EMBASE databases was carried out on papers published in English between 1966 and December 1999. Search terms were Parkinson(s) and hallucination(s)/and “visual”. Case series, surveys, and case-control studies were included, but not case reports and reviews. The minimum quality criterion was some information on how the diagnosis of Parkinson's disease was reached and qualitative information on abnormal visual phenomena. These criteria excluded most studies including those before 1990 (thoroughly reviewed by Cummings.3 Table 1 summaries the results of included studies published after 1991.

Prevalence

Prevalence rates of 11/89 (12.3%) for visual illusions plus hallucinations,12 or 23/189 (11.6%) for hallucinations13 are typical of samples of chronically treated patients. Recent surveys give prevalence estimates of visual hallucinations between 8.8% and 44%.14-19 The visual phenomena range from bizarre, complex, and frightening “visions”, through distortions of real percepts (illusions) to vague feelings of a “presence”. Aarsland et al 20 carried out a community based study of 235 patients in Norway with Parkinson's disease. Of these, 23 patients (9.8%) had hallucinations with retained insight, and another 14 patients (6%) had more severe hallucinations and delusions. Underreporting is a possible problem,15 even in interview studies, possibly because patients may fear being labelled as “mad”.

Risk factors

The review—based on a total of 316 hallucinators and 806 comparison subjects— showed consistent results in terms of risk factors for visual hallucinations (table 1). Some features such as visual impairment have been found to be associated with visual hallucinations in general21 and in other neurological disorders.22 Primary deficits in visual processing are associated with Parkinson's disease and have been exhaustively reviewed elsewhere.23 The possible link with sleep disturbances has also been discussed at length.11 24

MEDICATION

Comparisons of hallucinators and non-hallucinators have seldom shown major differences in drug history.16 Anticholinergic agents may be responsible for confusion in elderly people25 and cholinergic26 and serotonergic systems together have been implicated in hallucinations.27Some authors noted increased use of anticholinergic drugs (and primary dopamine agonists) but the results are inconsistent. Goetzet al 28 studied five non-demented patients with Parkinson's disease with daily hallucinations who were given high dose levodopa infusions in a placebo controlled trial. No hallucinations were provoked. Fernandezet al 29 found that when there was an apparent link between hallucinations and medication or clinical state it was more likely to be in periods of immobility. Hence visual hallucinations do not simply relate to high levels of dopaminergic stimulation but clearly dopaminergic drugs interact with Parkinson's disease in some way to produce visual hallucinations.

COGNITIVE IMPAIRMENT

Although some studies excluded patients with significant dementia,24 all samples with a range of cognitive impairment related to Parkinson's disease have shown a significant correlation with visual hallucinations (more impairment in hallucinators; table 1). Several mechanisms could account for the association, excluding potential confounders such as medication dose, duration of illness, and depression:

(1) Cognitive impairment may be a marker of diffuse brain disease or more severe basal ganglia pathology. However, a recent study reported that despite having more severe disease, hallucinating patients with Parkinson's disease did not show more evidence of white matter lesions on MRI30 and this concurs with earlier CT findings.31

(2) Cognitive impairment could also be a marker of a general degradation in information processing. A range of cognitive functions are impaired in Parkinson's disease.32 Visual hallucinations occur in many neuropsychiatric conditions as though at the end of a final common path.11 33 34 This could occur through combined neurological and psychological mechanisms.

(3) Finally, the impairment (or pattern of spared and compromised functions) may include abnormalities in, for example, reality testing, source monitoring, visual cognition etc, which are necessary to produce visual hallucinations. Waterfall and Crowe35systematically reviewed the neuropsychological literature on visual cognition in Parkinson's disease and concluded that complex visuospatial functions were compromised.36 Whether there is a pattern of cognitive deficits specific to visual hallucinators awaits further research.

AGE AND DURATION OF ILLNESS

Increased age has been associated with the presence of hallucinations.37 38 This might be explained by accerelated sensory loss39 40 or age related side effects of medication.41 One of the main confounders with age is duration of illness; when the non-independence of these variables was controlled, Fénelon et al 19found that duration of illness was the crucial factor. Grahamet al 17 identified two subgroups of patients with Parkinson's disease experiencing hallucinosis: in those with disease duration of 5 years or less, visual hallucinations were associated with rapid progression of the motor but not the cognitive component of the disease. In the remainder with longer histories, visual hallucinations were associated with postural instability, global cognitive impairment, and the lack of depression. Goetz et al 42 contrasted patients with Parkinson's disease who experienced hallucinations (visual and auditory) within 3 months of levodopa therapy with those who experienced hallucinations after 1 year of treatment. Diagnoses in the early onset group more often changed to Lewy body or Alzheimer's disease. Lewy bodies are present to a greater or lesser degree in all cases of Parkinson's disease43 and are known to be associated with visual hallucinations.44 45

Age of onset, duration, or speed of progression of illness, cognitive decline, and suboptimal response to medication are all proxy measures for disease severity. Most authors who recorded disease stage using the Hoehn and Yahr classification46 found a relation between severity and presence of visual hallucinations.

Finally, personality profiles as measured by the Minnesota multiphasic personality inventory (MMPI) have been implicated in hallucinations in patients with Parkinson's disease.47 48 A recent study controlled for cognitive impairment and found an MMPI pattern specific for hallucinations.14

After this review, we screened a community sample of patients with Parkinson's disease for the presence of current hallucinatory experiences. Patients who consented were then interviewed to gather more detailed information about their disorder and the characteristics of the hallucinations.

Results

The two groups differed significantly on duration of illness (t (42)=2.34, p<0.05), but did not differ in age (t (42)=−1.615, p>0.05), or levodopa medication (t (42)=1.887, p>0.05). A brief cognitive screening was performed which showed generally poorer performance in the hallucinators with a significant difference on a recognition memory test for faces. The hallucinators had greater disease severity (as assessed by the Hoehn and Yahr scale46) but did not differ on the MMSE. Hence, the clinical profile of this group was similar to others in which visual hallucinations have been studied. Brief details of hallucinatory experiences are given for each patient in table 3. Most described complex biological forms (animals, children etc) sometimes smaller than in real life and distorted. More fragmentary formless precepts were also noted.

The percentages of each group who reported the presence of problems or symptoms are shown in table 4. A χ² test was used to investigate whether the distribution differed from chance (table4).

FREQUENCY AND DURATION

Just over three quarters had five or less visual hallucinations a week (76.2%).All patients had their eyes open while experiencing the hallucinations. The duration of the experiences reported varied according to the time of day, with the longer episodes either in the morning or evening. In one patient, visual hallucinations lasted several hours and manifested gradually through the day. The remainder claimed that their hallucinations generally lasted for a few seconds up to 30 minutes and most (15 (76%)) experienced a sudden onset.

APPEARANCE

Seven patients experienced black and white images whereas 14 had single colour or multiple colour hallucinations. The clarity was blurred in 11 (52.4%) of the patients; only five patients (14.3%) described their hallucinations as sharp, although few reported distorted images. In these five patients the hallucinations were of meaningful content, pets or loved ones now dead, where perhaps familiarity may have contributed to the clarity. Thirteen patients (61%) claimed that their hallucination “seemed real” which could be taken as a measure of lack of insight into their hallucinatory nature. Sixteen (76%) reported having hallucinations that had form. Often the content was mundane such as an animal or unfamiliar person but ranged to the bizarre (faces made of brick). visual hallucinations involving people consisted of relatives, soldiers, and “strange people”, mainly involving children. Many of the hallucinations were dynamic with 18 patients (85.7%) reporting movement. Seven patients (33.3%), however, had recurrent hallucinations generally in the same surroundings; as one patient put it “like a video recording of an event, which replays every day”.

PRECIPITANTS AND CONTROLS

The hallucinations showed a clear relation with medication in only five patients (23.8%), who reported onset or aggravation of the symptoms in relation to consuming medication. The seven of 16 patients who claimed that their hallucinations were not linked to medication, experienced them in their immobile periods, usually at night or first thing in the morning. Ambient lighting was an important factor individually, with the largest group (52.4%) reporting visual hallucinations only in dim lighting. Fourteen (66.6%) described their experiences as involuntary. The seven patients (33.3%) who had control of their hallucinations were only able to terminate the image. The most common way of interacting with the hallucination was either by walking towards it or by trying to touch it. Twelve patients (57%) reported that the hallucination started with a percept that slowly changed into another image, and the remainder stated that the images appeared independently. Emotional responses at interview were varied, often in line with the severity of the hallucinations experienced: frustration, anger, fear, and in a few patients, indifference.

Discussion

Our sample may not be wholly representative of patients with Parkinson's disease and our case ascertainment methods may have resulted in selection biases for both hallucinators and non-hallucinators. Nevertheless, we were able to recruit a group of patients with hallucinations who were willing and able to describe their experiences, which was large enough to discern some general patterns. Previous studies based on hospital clinic attenders may be biased toward patients with more complex needs. Those patients with hallucinations differed from non-hallucinators for duration and severity of illness, as anticipated from the review of the literature. There was some evidence of cognitive impairment although the groups did not differ on the MMSE, probably because we sought only patients who could give a good account of hallucinations. More details of the neuropsychological assessment will be presented elsewhere.

This investigation indicated that hallucinations associated with Parkinson's disease possess a common set of characteristics. The typical hallucinator's experience occurs while alert and with eyes open, generally in dim surroundings. It involves having a blurry image appear suddenly without voluntary effort, filling an area of the visual field. The hallucination is present for a few seconds, typically moves while present, and then suddenly vanishes. The findings from this study concur with previous investigations. The visual hallucinations most often reported were complex, usually containing animate or inanimate objects or persons15 although more transient and less clearly perceptual phenomena also occurred.19 Usually they contained five or less images, which were sometimes meaningful to the patient. For example, one man regularly saw his dog, which had died 5 years earlier, lying by the bed. In addition, most reported that their hallucinations often occurred in dim surroundings, were non-threatening, and were sometimes recurrent.24 These characteristics are very similar to those documented in neurologically normal people with visual impairment (Charles Bonnet's syndrome49). A survey of 60 such patients showed that frequent meaningful complex hallucinations were the most common type and lasted seconds or minutes. As in the patients with Parkinson's disease, most occurred with the eyes open and in dim light, and were not stereotyped. Some slight differences also emerged in that the patients in this sample only experienced hallucinations with eyes open (as did most visually impaired people) and more often described the images as blurry and moving. Different patterns have been described in the visually impaired in relation to colour, duration, motion, and eye opening.50 Nevertheless, the similarities seemed to outweigh differences. In fact visual hallucinations in elderly people,21 those with Alzheimer's disease, and those with late paraphrenia51 or even typical schizophrenic psychoses52 also tend to be complex, brief, and associated with some sensory impairment and variable insight. Patients with Lewy body dementia show a divergence, with more auditory or multimodal experiences, and less insight.45 53

Looking at how the properties of hallucinations found in this study differ from those of normal visual perception, the most obvious is that visual hallucinations are only present for a brief time and are generally indistinct. The hallucinations also seem particularly susceptible to being identified as hallucinations when they are actively examined, at which point they tend to disappear. Even though some of the images appeared real, seemed to occur externally, and were not under voluntary control, most patients “knew” that they were hallucinating. Again, a similar pattern is seen in Charles Bonnet's syndrome.49 54 Consistent triggering and stereotyped form may have aided this judgement but were not frequent enough to provide a reliable basis for reality testing. The fact that some patients attempted to interact with the hallucination suggests that they were regarded with some suspicion. Hence insight tends to be preserved in the sense that the perceptual experience is relabelled as pathological,55 which contrasts with patients—not present in this sample—who develop delusions revolving around their experiences. The precise cognitive basis for accurate judgements of this type requires further study.

Visual hallucinations in the patient with Parkinson's disease probably result from a complex interaction of many variables, including cognitive, affective, medication, sensory, and even personality and environmental factors. The fact that these experiences can occur with little or no visual information from the environment (or none in the case of the blind) argues strongly for endogenous processes playing a part in the production of the hallucinations. These may be understood mechanistically as the manifestation of aberrant possibly disinhibited discharges from neural systems involved in processing or generating visual images.34 54 56 Indeed the apparent overlap in the phenomenology of hallucinations regardless of cause supports this idea. Similarly, some hitherto mysterious features such as micropsia and “tessellation” (brick-like patterns as seen in patients 7, 14, and 21) may be explained given a complex understanding of visual physiology.54 The predominance of human faces and dynamic forms may reflect the survival implications of these stimuli and subsequent evolution of specialised neural substrates for their processing.57-59 However, certain aspects of the content—the predominance of remembered people or animals, some of emotional significance—suggests that higher level “top down” cognitive factors may also play a part. Neisser60 argues convincingly that perception is a cyclic process involving receiving information from the environment through cognitive schemata, which in turn direct exploration for new information. Nevertheless the commonalities between the form of hallucinations described in treated patients with Parkinson's disease and other conditions suggest that visual hallucinations represent a restricted set of such schemata that are played out on a dysfunctional physiological system.