Visual hallucinations (VH) and disturbances of visual perception occur in a number of neurodegenerative diseases, and particularly in Parkinson’s disease (PD) and dementia with Lewy bodies (DLB). In PD, the prevalence of VH is between 6% and 60%1 and is known to be related to the distribution of Lewy body pathology.2 3 VH in non-Lewy body bradykinetic rigid syndromes has not been systematically studied but they have been reported in multiple system atrophy (MSA), progressive supranuclear palsy (PSP) and corticobasal degeneration.4^–7

In a pathological series of patients with parkinsonism, we have shown that VH are highly specific for Lewy body pathology and rarely occur in PSP and MSA.8 This led to the suggestion that VH might be helpful in the differential diagnosis of PD from atypical parkinsonism and the 5% of patients that have unclassifiable parkinsonism.9

Existing validated questionnaires used to screen for VH in PD, including the Unified Parkinson’s Disease Rating Scale (UPDRS)10 and the Neuropsychiatric Inventory (NPI),11 probably underestimate the frequency of VH in PD.12 13 We have, therefore, developed a simple and short inventory capable of rapidly screening for VH, including minor forms, to determine the incidence of VH in an outpatient population with bradykinetic rigid syndromes.

PATIENTS AND METHODS

Development of Queen Square Visual Hallucination Inventory

On the basis of Fenelon’s description of VH in PD, the Queen Square Visual Hallucination Inventory (QSVHI) was developed to determine the presence of minor and major visual hallucinations and illusions (see appendix).13 The questionnaire was designed to be read verbatim to patients in the context of a busy outpatient clinic. The screening question (question A) was equivalent to the “gold standard” questions used in current validated and operational inventories such as the UPDRS and NPI.10 11 Question B asks specifically for minor hallucinations: “presence” hallucinations; “passage” hallucinations; and illusions.13 Question C clarifies the screening question to exclude visual experiences other than hallucinations and if they occurred in isolation or were associated with auditory hallucinations. The physician was asked in question E to determine whether or not the hallucinations were related only to medications (VH started with initiation of medication and/or permanently disappeared on withdrawal of offending medication) or if the hallucinations occurred in the context of delirium.

Recorded data

The QSVHI was tested in consecutive patients in three movement disorders clinics in London (National Hospital for Neurology and Neurosurgery and University College London Hospital). The study was approved by the local research ethics committee. UPDRS parts II and III in the “on” state, the Mini-Mental State Examination (MMSE)14 and medication history were recorded, including calculation of a daily L-dopa equivalent unit dose15 16 The presence of depression, use of glasses and ocular pathology, including cataracts, retinal pathology and glaucoma was recorded.

Patients

A total of 181 patients were included: 117 (64%) patients met the UKPDSBB clinical diagnostic criteria for PD,17 five (2.8%) had DLB,18 23 (12.7%) met the NINDS-SPSP criteria for possible or probable PSP (three PSP-P and 19 Ricardson's syndrome),19 nine (5%) met the consensus criteria for MSA (three MSA-P and six MSA-C),20 five (2.8%) had probable vascular parkinsonism (VP),21 one had corticobasal degeneration, one had a mitochondrial cytopathy, one had orthostatic tremor and in 19 (9.5%) a definitive diagnosis had not been reached. This group of patients with unclassified or undetermined parkinsonism (UP) had either not satisfied any criteria for diagnosis or had overlapping features suggestive of more than one nosological entity. Fifteen consecutive disease control patients with non-neurodegenerative neurological conditions selected from a different clinic for patients without parkinsonism (idiopathic cervical dystonia, n = 7; hemifacial spasm, n = 1; dopa responsive dystonia, n = 1; essential tremor, n = 6), and 14 control subjects, without neurological disease, were also included after written informed consent was obtained. To determine sensitivity and specificity of VH, data from consecutive patients with a definitive diagnosis in the movement disorders clinic were used in a 2×2 table. This was also calculated for PD versus PSP.

RESULTS

The mean age of all patients was 66.8 (SD 11.7) years (range 20–100); 63% were men, and mean disease duration was 9.3 (SD 5.5) years (0.3–55) (table 1). In PD, L-dopa and dopamine agonists were used more frequently than in non-PD conditions, but compared with PSP there was no significant difference in the use of amantadine or anticholinergic medications. MMSE and Hoehn and Yahr scores were significantly lower in PSP than in PD, but UPDRS II and III scores were comparable. There were no significant differences between groups for retinal or refractory visual problems, depression or the frequency of anticholinergic, selegiline, cholinesterase inhibitor or antipsychotic medication use.

The screening question (QSVHI question A1) identified only 43 (38%) patients with clinically diagnosed PD who had VH, and none with PSP, MSA, VP or essential tremor. Two (14%) control patients answered yes to this question, describing what transpired to represent visual auras associated with migraine. Eighty-five patients (75%) with PD, three patients (20%) with VP and three patients (14%) with PSP reported minor hallucinations and illusions (QSVHI question B). Among those with UP, nine patients (47%) experienced visual hallucinations. Isolated auditory hallucinations occurred in 16% of patients with PD and in none of the other movement disorders. VH were deemed to be related only to medications in 3% (two patients) of the PD hallucinators, but were the cause in two of three of PSP hallucinators. Delirium was associated with VH in 7% (five patients) of PD hallucinators.

De novo VH, defined as VH not related to delirium or medications, were present in 78 (68%) patients with PD but only in one (5%) patient with PSP. That patient reported minor presence hallucinations that had been lifelong. Two patients with PD reported similar paranormal experiences going back to childhood. Two (22%) disease controls and one (7%) control reported presence hallucinations, limited to the months following the death of their spouse. In disease controls, the majority of VH were related to previous anticholinergic medication

Medication use and MMSE scores were not significantly different between PD hallucinators and non-hallucinators (27.5 vs 28.2; t test, p = 0.32), but PD hallucinators had significantly worse UPDRS II (mane 14 vs 10.7, t test, p = 0.02), UPDRS III (mean 30.6 vs 24.7, t test, p = 0.04), Hoehn and Yahr (2.6 vs 2.2, t test, p = 0.01) scores and longer disease duration (11.3 vs 6.6 years, t test, p<0.001).

The sensitivity and specificity of a positive response to individual QSVHI questions are summarised in table 2. In patients where the diagnostic dilemma was between PD and PSP, the specificity of de novo VH was 95% and the positive predictive value (PPV) approached 100%.

DISCUSSION

The QSVHI is a sensitive and simple method for establishing the presence of minor and major hallucinations in parkinsonism. In contrast with the UPDRS and NPI, the QSVHI prescriptively asks for minor visual hallucinations and illusions, and so doubles the sensitivity of standard clinical screening questions that have previously been accepted as the gold standard.10 11 Furthermore, the QSVHI is brief and can be more easily used in routine clinical practice than other rating scales.

VH, as assessed by the QSVHI, appear to be specific for the clinical diagnosis of PD (91% specificity). The PPV for the clinical diagnosis of PD has previously been determined as 98.6% by examination of pathological material1 22 and, taken together, these data imply that patients who experienced VH according to the QSVHI have a 90% chance of having underlying Lewy body pathology. This supports the findings of our retrospective clinicopathological study, where VH only exceptionally occurred in PSP and MSA. Others have also found that while PD hallucinators have more severe motor symptoms, medication use does not differ from non-hallucinators.7 13 23^–25 We and others have found that MMSE scores do not differ between these groups but the insensitivity of the MMSE for cognitive dysfunction in PD means that unmeasured cognitive differences cannot be excluded.

Fenelon’s minor hallucinations and illusions are not examined in detail in the visual hallucinations components of the UPDRS10 and NPI11 or other published questionnaires, including the Baylor PD Hallucination Questionnaire 26 and the Parkinson Psychosis Questionnaire.13 27 Minor or “benign” hallucinations and illusions form part of the early spectrum of disorders of perception that eventually lead to formed VH.12 28 The QSVHI was designed to embrace the clinical observation that minor hallucinations and illusions occur frequently in PD but are not spontaneously reported13 and may be missed using only screening questions used in the UPDRS or NPI (QSVHI question A). This finding may prove particularly helpful in the UP group where the screening question identified VH in 21%, but the complete QSVHI found 47% of UP patients experienced some sort of VH. These patients most likely had Lewy body pathology underlying their parkinsonism. The impact of the QSVHI on differential diagnosis is likely to be significant in the 4–5% of patients attending specialist clinics whose parkinsonism remains unclassifiable even after years of disease.9 Because VH in PD do not develop until late in the disease,8 screening for their presence is likely to be most helpful in patients with established but undifferentiated parkinsonism. The QSVHI does not replace detailed clinical evaluation of the patient but questions B and C can be included easily in the routine assessment of outpatients.

One of the potential weaknesses of this study is the selection bias inherent in a movement disorders clinic from a tertiary referral centre. The community prevalence of PD is estimated to be more than 20 times higher than PSP and MSA,29 30 so PD patients are probably under represented in this study. If more PD patients were included, the PPV in this study would have been increased. The lack of pathological confirmation also limits our conclusions, but the historical diagnostic accuracy in this clinic is known, so reasonable assumptions can be made.31

We would like to propose that VH should be added to the list of supportive findings (red flags) in step 3 of the UK Queen Square Brain Bank criteria on the basis of this study and our previous clinicopathological report, and that parts B and C of the QSVHI are useful for exposing the presence of visual hallucinations in PD which may otherwise go unrecognised.