J Neurol Neurosurg Psychiatry 84:505-510 doi:10.1136/jnnp-2012-304095
  • Cognition
  • Research paper

Clinical and cognitive correlates of visual hallucinations in dementia with Lewy bodies

  1. Mauro Dam2
  1. 1Department of Neurosciences, SNPSRR, University of Padova, Padova, Italy
  2. 2IRCCS San Camillo Hospital Foundation, Venice, Italy
  3. 3Department of Surgical and Gastroenterological Sciences, University of Padova, Padova, Italy
  1. Correspondence to Professor A Cagnin, Department of Neurosciences, Sciences NPSRR, University of Padova Medical School, Via Giustiniani 5, Padova 35128, Italy; annachiara.cagnin{at}
  • Received 6 September 2012
  • Revised 13 November 2012
  • Accepted 30 November 2012
  • Published Online First 21 December 2012


Background The presence of recurrent complex visual hallucinations (VHs) is a core feature of dementia with Lewy bodies (DLB). The aim of this study was to investigate which clinical and neuropsychological characteristics are associated with VHs and their predictive value over a 1 year follow-up.

Methods 81 DLB patients, 41 with (VH+) and 36 without (VH−) VHs, and 45 patients with Alzheimer's disease (AD), were enrolled. All participants underwent extensive neuropsychological testing. Visual–spatial and perceptual abilities were evaluated with the Visual and Object Space Perception (VOSP) battery. Fluctuations in attention, rapid eye movement sleep behaviour disorder (RBD) symptoms, extrapyramidal signs and behavioural disturbances were studied with dedicated clinical scales.

Results The presence of VHs was associated with older age and later disease onset, but not with disease duration or with fluctuations, RBD or parkinsonism severity. Cognitive correlates of VHs were deficits in visual attention (digit cancellation: p<0.005) and executive functions (clock drawing: p<0.05; digit span forward: p<0.05) on a background of a slightly worse global cognitive performance (Mini-Mental State Examination: p=0.05). Visual–perceptual and visual–spatial deficits were significantly worse in DLB than in AD patients (VOSP subtests scores 1, 6, 7 and 8) but were not different in DLB VH+ and VH−, except for subtest 6. Poor performance in the visual attention task was an independent predictor of VHs.

Discussion Impairment of visual–spatial and perceptual abilities in DLB represents a disease related cognitive signature, independent of the presence of VHs, for which it may represent a predisposing condition. Visual attention, instead, is the main cognitive determinant for the genesis of VHs.


Dementia with Lewy bodies (DLB) is the most frequent form of dementia after Alzheimer's disease (AD), accounting for 25% of all cases of dementia in the elderly. The pattern of cognitive decline in DLB is different from that in AD, and is typically characterised by attentive, executive and visual–perceptual deficits.1 ,2 Episodic memory, instead, is frequently preserved in the early stage of the disease. Visual hallucinations (VHs), fluctuations in cognition and alertness, and parkinsonism represent the core features of the diagnostic criteria for DLB.3 The presence of VHs at presentation yields the stronger positive predictive value for the diagnosis of DLB with respect to the other core features.4 VHs consist of recurrent and well formed complex vivid images, may be present from an early stage of the disease and are associated with poor functional outcome. Patients often have lack of insight about the hallucinatory nature of the event that may represent a cause of intense emotional distress. The brains of DLB patients with VHs have a higher burden of Lewy body pathology in the inferior temporal and parahippocampal cortex, and in the amygdala.5

The cognitive correlates of VHs have been poorly investigated in DLB. Some possible insights come from studies in patients with Parkinson's disease (PD). In a recent study on the aetiological model of VHs in PD, independent determinants for the presence of VHs were: (a) frontal cognitive deficits, (b) visual–perceptive alteration, (c) symptoms of rapid eye movement (REM) sleep behaviour disorder (RBD) and (d) autonomic dysfunction.6 A relationship between the development of VHs and ocular pathology has not been found, thus attributing a more stringent role for the emergence of VHs to the impairment of central visual processing than to a defect in peripheral visual acuity. These findings are in line with the Perception and Attention Deficit model for recurrent complex VHs, which requires the coexistence of attentional (dorsolateral frontal cortex) impairment and visual–perceptual (ventral visual stream) deficits.7 The combination of impaired attentional binding (top down functions) and defective visual sensory processing (bottom up functions) may favour a reduced gate control on the emergence of internal false images, ultimately responsible for VHs.

In this study, we evaluated which clinical and neuropsychological characteristics were associated with the presence of VHs in DLB patients and whether they could predict the occurrence of VHs over a follow-up period of 1 year.



Patients were selected among those referred to the outpatient memory clinic at the Department of Neurosciences, Padova, Italy. The diagnosis of DLB was made according to the consensus criteria recommended by the Consortium on DLB.3 The diagnosis of probable AD was based on the criteria of the National Institute of Neurological and Communicative Disorders and Stroke, and the AD and Related Disorders Association.8 Exclusion criteria included evidence of severe cerebrovascular disease, as assessed by brain CT or MRI scan; a severe degree of global cognitive impairment, as expressed by a Mini-Mental State Examination (MMSE) test9 score <10/30 (as it would have impaired the possibility of sustaining intensive neuropsychological testing); a severe eye pathology that impaired visual acuity; a previous history of psychiatric disorders; and the absence of a caregiver who could provide valid information about the patient's behavioural disturbances or cognitive fluctuations. The only antiparkinsonian medication allowed was levodopa. Patients on cholinesterase inhibitors (ChEI) were eligible for the study if they were on a stable dose for at least 3 months.

We studied 126 patients (mean age 75±6.4 years (range 55–86)). Among these patients, 81 (64.3%) had a diagnosis of probable DLB and 45 (35.7%) received a diagnosis of probable AD.

The study conformed to the guidelines set out in the Declaration of Helsinki of 1975 and the study received approval from the institution review board. All subjects provided written informed consent.

Clinical evaluation

Each subject underwent a full clinical history and neurological evaluation. Clinical data collection included revision of records with drug treatments taken in the previous 6 months. Extensive general and neurological examinations were undertaken, with detection of arterial blood pressure, heart pulse rate and visual acuity to rule out severe eye pathology (cataract, glaucoma, macular degeneration) which could be responsible for impairment of visual abilities. In particular, the neuro-ophthalmologic assessment included the external inspection of the eyes, pupil reactions, penlight reflex, measure of near vision acuity, ocular movements and estimation of the visual field by confrontation test. The ocular fundus was assessed by direct ophthalmoscopy. A brain CT or MRI scan was performed for each subject to assess the degree of vascular brain changes.

Patients were screened for extrapyramidal signs using the Unified Parkinson's Disease Rating Scale motor score (part III).10 Information on hallucinatory episodes was obtained from each patient and/or related caregiver at the time of cognitive evaluation. The presence of VHs was also ascertained from the clinical records when available. The frequency and severity of VHs in the past 4 weeks were assessed using the Neuropsychiatric Inventory (NPI) questionnaire.11 Only patients with recurrent complex VHs were included in the study, while visual illusions, feeling of presence and passage were not considered.

Fluctuations in cognition and alertness were assessed with the Mayo Fluctuations Questionnaire administered to the informant.12 The presence of RBD symptoms was determined using the Mayo Sleep Questionnaire completed by the informant.13

Clinical investigations were performed by neurologists specialised in old age neurology in a clinical setting and in the presence of a caregiver. All DLB patients included in the study underwent neurological follow-up visits for at least 12 months after the first neurological and neuropsychological assessments to confirm the clinical diagnosis and to detect the occurrence of VHs in this follow-up period.

Neuropsychological assessment

All patients underwent an extensive neuropsychological evaluation. Global cognitive performance was assessed using the MMSE test.9 For the assessment of specific cognitive domains, the following neuropsychological tests were administered: digit cancellation (visual search)14 and trail making test—part A15 for evaluation of attention; digit span (forward and backward)16 and prose memory (immediate and delayed recall) test15 for evaluation of short and long term memory, respectively; letter fluency test15 for the study of ideational capacities and executive functioning; clock drawing test15 for the evaluation of visual–constructional abilities and executive functions; and the Rey–Osterrieth Complex Figure (ROCF) test for assessment of visual–constructional abilities (ROCF copy) and long term visual–spatial memory (ROCF delayed recall).17

The Visual and Object Space Perception (VOSP) battery was administered to all patients to assess visual–perceptual and visual–spatial impairments.18 VOSP is a neuropsychological battery composed by a screening test, four subtests for object perception and four subtests for space perception.

The screening test evaluates figure/ground perception. It is composed by 20 squares with random patterns, half of which have a degraded ‘X’ inserted. The subject is asked to judge whether the ‘X’ is present or not (maximum score 20).

For object perception, in the incomplete letters task (subtest 1), the subject has to recognise 20 degraded capital letters (maximum score=20); in the silhouettes task (subtest 2), the subject is asked to identify 30 black silhouettes presented in an unusual view (maximum score=30); in the object decision task (subtest 3), the subject has to judge which of the four drawings represents a real object, while the three other objects are nonsense drawings (maximum score=20); and in the progressive silhouettes task (subtest 4), two series of 10 drawings progressively easier to identify were presented (maximum score=20).

For space perception, the dot counting subtest (subtest 5) is composed by 10 arrays of black dots and the subject has to count the number of dots in every item (maximum score=10); in the position discrimination task (subtest 6), two squares are presented, one with a black dot printed in the centre and the other with a black dot printed out of the centre: the subject has to point to the dot perceived in the centre of the square (maximum sore=20); in the number location task (subtest 7), each item consists of two squares, one containing a random pattern of numbers and one containing a single black dot, corresponding to the position of one of the numbers: the subject has to point to the number that corresponds to the position of the dot (maximum score=10); and in the cube analysis task (subtest 8), the subject has to count how many solid bricks there are in a series of drawings, each representing a tridimensional structure of bricks (maximum score=10).

Statistical analysis

The t test for independent sample, Mann–Whitney U test and χ2 test were used for normal, ordinal and categorical variables, respectively. A logistic regression model was performed to find the independent predictors of hallucinations in the DLB group. The significance level was set at p<0.05. Bonferroni correction for multiple comparisons was also applied to the neuropsychological test scores, with a significance level set at p≤0.012.


Clinical findings

Demographic and clinical characteristics of the DLB and AD groups are summarised in table 1. The DLB group was also divided in two subgroups on the basis of the presence or absence of VHs (VH+ and VH−, respectively).

Table 1

Demographic and clinical characteristics of all of the subjects enrolled in the study

The DLB and AD groups were matched for age, gender and level of education. Disease duration and age of disease onset were similar in the two groups. The presence of VHs was reported in 45 of 81 patients with DLB (55.6%) whereas it was not detected in AD patients (p<0.001). None of the DLB VH+ patients was experiencing VHs at the time of the neurological and neuropsychological examinations. The context of VHs was most often characterised by well formed figures of people or distorted human faces, more rarely by animals. Insight about the hallucinatory nature of the episode was lacking in 87% of VH+ patients. Thirteen of 45 VH+ patients experienced fearful delusions associated with VHs. One patient had concomitant visual and auditory hallucinations. The NPI subscores related to the VH item for the DLB VH+ group are shown in table 1. NPI VH subscore frequency was 2.5±1.1 (severity 2.1±0.8).

Within the DLB group, VH+ patients were older and with a later disease onset than VH− patients. Disease duration was similar between DLB VH+ and DLB VH− patients (p=0.53).

Moreover, VH+ patients had a higher frequency of RBD symptoms than VH− patients, with a trend towards statistical significance (p=0.07), and a greater burden of behavioural disturbances, as shown by the higher NPI total score (p=0.01). However, when considering the NPI total score, minus the NPI subscore related to the VH item, the burden of remaining behavioural disturbances was similar between the VH+ and VH− groups (p=0.76). No associations between VHs and the Unified Parkinson's Disease Rating Scale motor score or fluctuations in attention (p=0.59) were detected. In the longitudinal study, six of 36 DLB VH− patients developed VHs during the 12 month follow-up period.

Fluctuations in attention and RBD symptoms were not associated with disease duration (p=0.44; p=0.23), patient age (p=0.96; p=0.65) or age at disease onset (p=0.95; p=0.59). There was also no association with the total NPI score.

Regarding drug treatments, only six of 81 patients with DLB (four VH+ and two VH− patients) were taking low dose levodopa (mean dose: VH+ 300±122 mg/day, VH− 225±105 mg/day). Given this small number of subjects, statistical analysis was not made and levodopa treatment was considered to have no influence on VHs.

Antipsychotics were used almost exclusively in VH+ DLB patients (VH+ 36%, VH− 3%; p<0.001). The only drugs used were quetiapine and clozapine at low doses (mean dose 39.5±26 mg/day).

In contrast, ChEI drugs were used in a similar proportion of DLB VH+ and DLB VH− patients at the time of the neuropsychological assessment (VH+ 47%, VH− 33%;  p=0.2), and all were on stable therapeutic doses. The mean NPI subscore of the VH item was not associated with ChEI use (mean NPI-VH score: ChEI+=5.5±4.4; ChEI−=5.7±4.2; p=0.9).

Neuropsychological findings

Neuropsychological tests results of AD and DLB patients are reported in table 2. The mean MMSE score was not different between the DLB and AD groups (p=0.57). Patients with DLB showed greater impairment of attention and visual–constructional abilities, performing worse than AD patients in the trail making test—part A (p=0.02) and in the ROCF copy test (p=0.01). For visual–perceptual and visual–spatial abilities, the DLB group performed worse than the AD group in all VOSP subtests, with significant differences detected in the incomplete letters subtest (p<0.005), position discrimination (p=0.03), number location (p=0.012) and cube analysis (p<0.001). In contrast, AD patients showed greater impairment in tests exploring memory, obtaining lower scores than DLB in the prose memory immediate recall (p<0.01) and delayed recall (p<0.001) tests. After Bonferroni correction for multiple comparisons, the scores for the following tests remained significantly different: ROCF copy, prose immediate and delayed recall, VOSP incomplete letters, number location and cube analysis subtests.

Table 2

Neuropsychological test scores in patients with Alzheimer's disease and dementia with Lewy bodies

Differences in neuropsychological test results between DLB VH+ and DLB VH− patients are summarised in table 3. The DLB VH+ group obtained lower MMSE scores than the DLB VH− group, although the difference was weakly significant (p=0.05). The presence of VHs was associated with worse scores in the neuropsychological tests assessing attention and executive functions. In detail, the DLB VH+ group showed greater impairment compared with the DLB VH− group in the digit cancellation test (p<0.005), clock drawing test (p<0.05) and digit span forward test (p<0.05), assessing visual attention, constructive functions and short term memory, respectively. The performance in each VOSP subtest was similar between the DLB VH+ and VH− groups, with the exception of the position discrimination subtest in which the DLB VH+ group obtained a slightly worse score (p=0.05). Only the digit cancellation test remained significant after Bonferroni correction for multiple comparisons.

Table 3

Neuropsychological test scores in patients with dementia with Lewy bodies, with and without visual hallucinations

In the longitudinal 12 month follow-up study, the six DLB VH− patients who converted to VH+ had a baseline cognitive profile similar to that observed in the VH+ group. In particular, they performed similarly to the DLB VH+ group in those neuropsychological tests which better differentiate between VH+ and VH−, such as the digit cancellation test (mean score=27.7±10), clock drawing test (mean score=1.8±2) and the position discrimination VOSP subtest (mean score=15.7±4). In the analysis of the VOSP, when the six DLB patients who developed VHs in the follow-up study were pooled together with the VH+ patients at baseline, the position discrimination subtest became significantly different in this combined group from the VH− group (p=0.01).

In the step forward regression logistic analysis, the only independent predictor of VHs was visual attention deficit, assessed with the digit cancellation test (p=0.004).

Fluctuations in attention were associated with poor performance only on visual attention (digit cancellation, p<0.05), and not on MMSE (p=0.09). On the VOSP battery, patients with cognitive fluctuations showed a more severe impairment in the object decision subtest (p=0.01).

DLB patients with symptoms of RBD had a worse performance on the VOSP position discrimination subtest (p=0.007). Neuropsychological test scores assessing other cognitive domains were not significantly worse in patients with RBD.


In this study, the presence of VHs in DLB patients was associated with older age and later disease onset, but not with disease duration. Cognitive correlates of VHs were deficits in visual attention and executive functions, on a background of slightly worse global cognitive performance. The DLB group showed impaired processing of visual information, which represents a disease related cognitive signature independent of the presence of VHs. Therefore, impairment of visual–spatial and visual–perceptual abilities appears to be a predisposing condition, not sufficient for the genesis of VHs. This visual cognitive deficit was expressed as both an impaired perceptive identification of images (ventral temporo-occipital pathway function) and an altered spatial localisation of a visual stimulus (dorsal parieto-occipital pathway function).

These results fit with the cognitive pathogenetic model for recurrent complex VHs proposed by Collerton et al, the Perception and Attention Deficit model, which suggests that VHs are induced by a combination of impaired attentive binding and poor sensory visual activation.7 The coexistence of impairments of top down and bottom up functions decreases the gate control on the emergence of internally generated false images. These false images are ‘proto-objects’ stored as memory based images, which are evoked by distorted sensory information and may reach the conscious level when adequate top down control from attentional processes is lacking. Deficits in central visual sensory processing, further impaired by poor environmental light such as in the evening or night, and concomitant alterations in attentional binding, particularly when fluctuation in alertness and attention are present, contribute to the emergence on the surface of consciousness of internally generated images, giving rise to recurrent complex VHs. These concepts have also been integrated into a similar model of VHs in PD for which the presence of VHs is generated by dysregulation between the gating and filtering of external perceptions, such that secondary to abnormal processing of visual stimuli, and an internal image production, modulated by REM intrusion or dopaminergic stimulation of the mesolimbic regions.19 More recently, failure to integrate ventral and dorsal attentional networks with the default mode network has been said to have a role in the development of VHs.20

The cognitive correlates of VHs have been poorly investigated in DLB. Mosiman et al found that DLB and PD-dementia patients with VHs were more impaired in tasks assessing visual–perceptual abilities.21 However, the group size of DLB patients with VHs was small and the neuropsychological battery used was not tailored to detect differences in attentional or executive functions. Previous work has focused on hallucinated PD patients with and without dementia. Gallagher et al showed that the cognitive determinants of VH development in PD were frontal and visual–perceptual deficits.6 Although many studies have drawn attention to the importance of executive dysfunction in the development of VHs in PD,22–25 only a few studies have investigated attentional control mechanisms specifically and the cross talk between the impairments of top down and bottom up cognitive functions. There is evidence that impairment of object and space perception, in association with poor sustained visual attention, might play a role in the pathogenesis of VHs in PD.26 ,27 One study focusing on combined executive and visual–spatial deficits in patients with PD-dementia and VHs supported the hypothesis that visual attention deficits are stronger determinant of VHs than impairment of visual stimulus processing.28 Although the bulk of studies of cognitive correlates of VHs have been undertaken in PD patients, many clinical and neuropsychological differences may exist among PD and DLB. These differences relate to a different regional distribution of synuclein pathology, extent of neuronal loss and a temporospatial pattern of spreading of the pathological processes. In DLB, synuclein pathology is localised earlier and more diffusely than PD in cortical regions, often coexisting with amyloid plaque deposition, and in the midbrain cholinergic and monoaminergic nuclei.29 ,30 In this study, VHs were associated with older age and later disease onset, while no relationship was found with disease duration or severity of motor symptoms. This latter finding represents a substantial difference from that observed in PD, where the development of VHs has been found to be associated with patient age and disease duration, occurring with worsening of cognitive and motor symptoms.6 ,22 ,28 ,31

During the 12 month follow-up period, six DLB VH− patients developed VHs and became VH+. These patients showed a profile of cognitive impairment comparable with that observed at baseline in the VH+ group, and the main difference from the stable VH− patients was the performance in the digit cancellation test. We therefore suggest that poor performance on visual attention tasks might have the strongest predictive value for the genesis of VHs in DLB.

Another interesting observation comes from the relationship between the presence of RBD symptoms and VHs. RBD is included among the supportive features for the clinical diagnosis of DLB. It has been suggested that VHs may represent a narcolepsy-like state, with intrusion of eye movement dream imagery into wakefulness,32 although this hypothesis has not been confirmed.33 Nevertheless, in PD patients, the presence of VHs has often been found to be associated with a greater frequency of RBD.6 Noteworthy, in this study we observed that the DLB VH+ group reported more RBD related symptoms than the VH− group, although this association was weak. The presence of RBD, in turn, was associated with poor performance on the position discrimination subtest, the most impaired VOSP subtest in DLB VH+ patients. Therefore, performance on one of the most difficult visual–spatial tasks of the VOSP battery is associated with the presence of both RBD and VHs in DLB. We can speculate, therefore, that alterations in those brain structures responsible for the genesis of RBD, such as the dorsal mesopontine nuclei, may have a role in the impairment of visual–spatial information processing. The cholinergic pedunculopontine nucleus has direct connectivity with the visual thalamic nuclei, including the lateral geniculate nucleus and the superior colliculus. In this perspective, it is also intriguing to consider that ponto-geniculo-occipital waves appear during the REM phase of sleep. Therefore, alteration of the dorsal mesopontine nuclei could be responsible for disruption of ponto-geniculo-occipital afferent projections predisposing to RBD and altering functional connectivity in the occipital cortex.34 In addition, cholinergic neurones of the pedunculopontine nucleus project directly to the cerebral cortex, although the most important neuronal sources of cortical cholinergic projections are the nucleus basalis of Meynert in the basal forebrain.

Apart from deafferentation of the occipital cortex from the subcortical cholinergic nuclei, alterations in cortico-cortical connectivity also have a central role in a disease where widespread cortical pathological and neurochemical changes are present, such as DLB. In this context, the finding that decreased fractional anisotropy in the inferior longitudinal fasciculus (ILF), a measure of white matter tract abnormalities, is associated with VHs in DLB is of particular interest.35 In fact, the ILF carries the temporo-occipital projections, connecting the associative visual area with those temporal regions with the highest pathological burden in DLB VH+ patients.5 Moreover, the integrity of the superior and ILF is associated with a good performance on visual search tasks in healthy subjects.36 Therefore, cortico-cortical dysconnection of the occipital cortex from regions of the temporal lobe may play a role in the genesis of VHs and contribute to the deficit in visual attention.

The main strength of this study is the large sample size of DLB patients, well balanced for the presence of VHs, and matched with AD patients as control group. In fact, the use of another type of neurodegenerative dementia as a control group allowed for better discrimination between the cognitive deficits which are disease specific (visual–spatial/perceptive deficits as a marker of DLB) from those which are condition specific (visual attention deficit as a marker of VHs). There are, however, a few limitations of our study. Visual–spatial and visual–perceptual abilities were investigated with one battery of dedicated tests, the VOSP, in which some cognitive aspects, such as visual imagery, face discrimination and motion perception, are not included. On the other hand, the use of all VOSP subtests allows discrimination between a deficit of visual–constructive and executive functions and impairment of visual central processing. Finally, this is the only study in the literature that used all of the subtests of the VOSP, thus giving a wide understanding of which subtests may be more sensitive in detecting cognitive visual impairment specific to DLB pathology. In this context, impairment of visual–perceptual (incomplete letters) and visual–constructive (cube analysis) abilities can discriminate between DLB and AD patients, whereas impairment in a specific subtest assessing visual–spatial function (position discrimination) seems to be worse in DLB patients with hallucinations. It is possible that performance on this subtest, which is particularly difficult as it requires discrimination between two very similar items, might better correlate with disease severity, being affected when the visual processing alteration extends from the ventral visual stream to the dorsal visual pathway.


We would like to thank Nicola Bonetto and Sara Pompanin for patient referral and selection.


  • AC and FG contributed equally to this paper.

  • Contributors AC was responsible for the conception and organisation of the research project and wrote the manuscript. FG was responsible for data acquisition and organisation of the project and wrote the first draft. NJ contributed to data acquisition and analysis. SF was responsible for data acquisition and reviewed the manuscript. GZ contributed to the organisation of the project and data acquisition. ME did the statistical analysis. MD contributed to the conception of the study and critically reviewed the manuscript.

  • Funding None.

  • Competing interests None.

  • Ethics approval The study was approved by the institutional review board.

  • Provenance and peer review Not commissioned; externally peer reviewed.


Visit the full archive of podcasts for JNNP here >>

Free sample
This recent issue is free to all users to allow everyone the opportunity to see the full scope and typical content of JNNP.
View free sample issue >>

Don't forget to sign up for content alerts so you keep up to date with all the articles as they are published.