Amardeep Singh, MD Department of Ophthalmology Copenhagen University Hospital Roskilde DK-4000 Roskilde Denmark Email: firstname.lastname@example.org
Purpose: To investigate the prevalence and clinical characteristics of the Charles Bonnet Syndrome (CBS) in a group of Danish patients with neovascular age-related macular degeneration (AMD) and to study whether CBS is associated with a specific retinal morphology.
Methods: Three-hundred consecutive patients with neovascular AMD attending assessment consultations following variable series of ranibizumab therapy were actively asked whether they had symptoms of CBS. If they responded positively, a detailed questionnaire was orally administered to inquire into the details of the symptoms. Detailed optical coherence tomography and autofluorescence was performed. A comparison was made between retinal morphology of a randomly selected equal number of patients without CBS to patients with CBS.
Results: Twenty-five (8.3%) patients of 300 had hallucinations attributable to CBS. The median lesion size – measured as total area with increased autofluorescence – in the CBS group (median 14.2 mm2) was not significantly different from the non-CBS group (median 16.2 mm2); however, the patients with CBS had significantly larger areas of geographic atrophy (median 2 mm2) compared to patients without CBS (median 0.3 mm2) (p = 0.002).
Conclusion: CBS is not uncommon in an unselected population with neovascular AMD, and symptoms of CBS may be associated with larger areas of geographic atrophy.
The Charles Bonnet syndrome (CBS) is characterized by the presence of vivid, complex and recurrent visual hallucinations occurring in psychologically normal patients (Schadlu et al. 2009). Although best described in age-related macular degeneration (AMD), this phenomenon may occur in any condition causing vision loss (Shiraishi et al. 2004; Tan et al. 2004). The symptoms of CBS vary from seeing geometric figures to experiences of seeing people or animals. Even though the content of the hallucinations in CBS is gen-erally not distressing to the patient; they may, however, cause fear of impending insanity (Menon et al. 2003). The prevalence of CBS has been reported to vary from <1% to 40% in different ophthalmological populations, but it is likely that this variation is because of differences in inclusion criteria, inconsistent depth of questioning and reluctance of patients to admit to having halluci-nations (Schadlu et al. 2009). Furthermore, differences in cultural or ethnic groups and type of disease may also play a role in the variation. In fact, only about 1/5 of the patients admit to have told others about their symptoms (Vukicevic & Fitzmaurice 2008).
The purpose of this study was to investigate the prevalence of CBS in a Danish group of patients with neovascular AMD and to examine the clinical characteristics of CBS in this population. Furthermore, we wanted to investigate whether the symptoms are associated with a specific retinal morphology, such as loss of photoreceptors, areas of geographic atrophy and thickness of subfoveal fibrosis.
Materials and Methods
Three-hundred consecutive patients with neovascular AMD attending assessment consultations following a variable series of ranibizumab treatments were all actively asked if they had experienced visual hallucinations. They were asked directly using a question as proposed by Holroyd with slight language modifications: ‘Some patients, with similar eye problems as you, report seeing things which are not really there, or which other people do not see. Have you ever experienced this?’ (Holroyd et al.1992). The hallucinations had to be complex and recurrent so if the patients only described the presence of elementary hallucinations, such as floaters and flashes of light, they were not considered as having CBS. Also patients who had a history of psychiatric or neurological disease, such as schizophrenia, dementia and Parkinson’s disease, substance abuse or were taking medications known to cause visual hallucinations were excluded from having CBS. All patients who admitted to having visual hallucinations attributable to CBS underwent a semi-structured interview administered by researcher AS. Specific characteristics of the hallucinatory experiences, living situation, the presence of auditory hallucinations and a medical and psychiatric history as well as substance and medication use were all noted. At the end of the interview, the patients were briefly informed about the benign nature of their visual hallucinations. Informed consent was obtained from all participants.
Visual acuity was measured both with Snellen chart and with ETDRS chart. Autofluorescence and detailed Spectral Domain Optical Coherence Tomography (SD-OCT) were performed on Heidelberg Spectralis HRA-OCT™ (Heidelberg Engineering, Heidelberg, Germany). Images of patients with CBS as well as an equal number of randomly selected images of patients without CBS were graded to determine lesion size, size of areas with geographic atrophy, photoreceptor integrity, and degree of subretinal fibrosis. The total area of the lesion (defined as increased autofluorescence) as well as total areas of geographic atrophy (defined as complete loss of autofluorescence) was measured using the software algorithm Heidelberg Spectralis HRA-OCT™. The integrity of the photoreceptor interface was graded (Hayashi et al. 2009). Briefly, the line representing the inner segments and outer segments of retinal photoreceptors (IS/0S) was identified. Subsequently, the patients were divided into three different groups; one group being where the IS/OS line was detected as a complete line beneath the fovea, another group where the IS/OS line was detected as a discontinued line beneath the fovea and a third group being where the IS/OS line could not be detected beneath the fovea.
To investigate whether a specific retinal morphology is associated with CBS, we compared autofluorescence and detailed OCT scans of the patients with CBS to autofluorescence and OCT scans of 25 randomly selected patients without CBS. The area of autofluorescence was measured by TLS using Heidelberg Eye Explorer. The thickness of subretinal fibrosis was measured using software included in the Heidelberg Eye Explorer which enables you to move the boundary measurements freely for exact measuring of subretinal fibrosis thickness.
The grading was carried out in blinded fashion so that the person who graded the pictures (TLS) was unaware of whether the patient had CBS or not. Pictures were not gradable in four patients with CBS because of blurred media.
As the data were not normally distributed, data are given as median and range. Comparisons between two groups were performed with the Mann–Whitney test, and correlations were performed with Spearman rank correlations. A p-value of <0.05 was considered significant.
Prevalence and clinical characteristics
Thirty-one of 300 patients admitted having hallucinations. Upon further interview, we excluded six patients. In the excluded group, one patient reported having seen a cat, but refused to talk more about it; two patients reported seeing person-size dark shadows; one patient reported being able to see through objects; one patient had been hallucinating since youth and did not have insight into the unreal nature of those hallucinations; and one patient had seen a person before falling asleep, but this was a one-time experience. Many patients reported seeing shadows and floaters and they were not considered as having CBS even though two of these saw person-size dark and vivid shadows. Ultimately, 25 (8.3%) of the patients were categorized as having CBS. One of these patients suffered from multiple sclerosis and was being treated with Baclofen, a drug known to have hallucinatory effects upon withdrawal (D’Aleo et al. 2007). Although one case report links Baclofen treatment to the occurrence of CBS (Stofler et al. 2004), we chose to include this patient in our population of CBS.
The nature of the hallucinations is shown in Table 1. The content varied from distinct objects and/or persons – so the same patient could experience both geometric figures and more complex objects such as persons. Most commonly, the symptoms lasted seconds or minutes. The hallucinations predominantly occurred either daily or weekly, and approximately half the patients had experienced the hallucinations for more than 12 months.
Table 1. Characteristics of hallucinations.
* One patient could keep hallucinating as long as she wished to.
Time of occurrence
At all times
Thirty-six per cent of the patients had not told anyone else about the hallucinations. Sixty per cent of the patients were living alone. Half of the patients said that they have some hearing difficulties.
Sixty-eight per cent did not find the hallucinations distressful. However, one patient found CBS to be very distressful.
There were 164 women (59%) and 111 men (41%) with a median age of 80 years (range 48) in the non-CBS group and 18 women (70%) and seven men (30%) in the CBS group also with a median age of 80 years (range 28). We found more women (70%) in our group of patients with CBS compared to the patients without CBS (59%).
The visual acuity on the best seeing eye was in the CBS group (median 0.45 range 0.9) significantly lower compared to the non-CBS group (0.5; range 1.2) (p = 0.048; Mann–Whitney).
The median lesion size – measured as total area with increased autofluorescence - in the CBS group was 14.2 mm2 (range 37.6) and not significantly different from the non-CBS group median 16.2 mm2 (range 46). However, the patients with CBS had significantly larger areas of geographic atrophy – measured as areas with no autofluorescence (median 2 mm2; range 12) compared to the non-CBS groups (median 0.3 mm2; range 8) (p = 0.002; Mann–Whitney) (Table 2 and Fig. 1). We did not see any difference in subfoveal photoreceptor integrity and degree of subfoveal fibrosis between the two groups. Subretinal fibrosis did, however, correlate negatively with visual acuity (p = 0.005; ρ = −0.56; Spearman rank correlation). Even though total areas of geographic atrophy correlated significantly to IS/OS integrity (p = 0.04; ρ = −0.43; Spearman rank correlation), they did not correlate with visual acuity (data not shown).
Table 2. Areas of lesion size and geographic atrophy in the CBS vs non-CBS group.
Lesion size (mm2)
Geographic atrophy (mm2)
14.2 (range 37.6)
2 (range 12)
16.2 (range 46)
0.3 (range 8)
The aim of this study was to investigate the prevalence and clinical characteristics of CBS patients with neovascular AMD. Because treatment has become effective in this group of patients, the number of patients attending the treating eye departments has increased immensely; hence, a better understanding of the prevalence of CBS will increase health care professionals’ ability to care for these patients.
Our data show that CBS in this population is not rare. Twenty-five patients (8,3%) receiving treatment with anti-VEGF reported symptoms of CBS. This prevalence rate is lower compared to some previously reported rates (Khan et al. 2008), but close to the prevalence of 13% described in 100 consecutive patients with AMD (Holroyd et al. 1992) and 11% described in 300 low-vision patients (Teunisse et al. 1995). An Australian study of 200 patients with variable retinal disease and a visual acuity of 0.5 or less found a prevalence rate of 17.5% (Vukicevic & Fitzmaurice 2008), while another report of 66 patients with AMD and a visual acuity of logMAR 0.6 or less found that 15% of the patients reported symptoms of CBS (Abbott et al. 2007). If we exclude all patients with a visual acuity of more than 0.5 on the best seeing eye from our material, we get a prevalence rate of 12%, which increases to 15% if we only include patients with a visual acuity of 0.3 or less on the best seeing eye. The prevalence rate rises to 28% if we only include patients with a visual acuity of 0.1 or less on the best seeing eye, indicating that bilateral loss of vision is an important risk factor for the development of CBS, which has been suggested previously (Khan et al. 2008).
As the occurrence of CBS is related to visual acuity, some of the differences reported may be because of different selection criteria in the different studies. In our study, we asked 300 consecutive patients in a daily clinic, in our view, resembling an average population of patients with neovascular AMD being treated with anti-VEGF.
Female gender has, to a variable degree, been suggested to be a risk factor for CBS. In our study, there were more women in the CBS group compared to the non-CBS group. However, as there is a predominance of women in the participating group of patients, a specific gender effect is difficult to identify. Although increasing age has been implicated as a risk factor for the development of CBS in other studies (Teunisse et al. 1995), we did not find any association with age and symptoms of CBS.
The nature of the hallucinations varied from complex hallucinations such as persons and/or animals to geometric patterns. In our population, seven patients (28%) reported seeing people, which is comparable to 24% reported by Vukicevic and 20% reported by Khan (Vukicevic & Fitzmaurice 2008; Khan et al. 2008). Nine patients (36%) reported having daily hallucinations which is comparable to 34% as has been reported by Abbott and co-workers, but somewhat higher than reported in the Australian population, where 24% of the patients reported having daily hallucinations (Abbott et al. 2007; Vukicevic & Fitzmaurice 2008).
In many instances, patients are reluctant to tell people in their surroundings or health care professionals about their hallucinations. This hesitation is certainly a contributing factor to the underestimation of the prevalence of CBS. We found that nine patients (36%) had not told anyone about their hallucinations – a clear indication that many patients will not reveal their hallucinations unless confronted with a direct question posed in the correct manner. In the Australian population where they used the same initial screening question as we did (Vukicevic & Fitzmaurice 2008), they found that 21% had not told anyone of their hallucinations. Our and their finding seems to suggest that many patients will admit to having hallucinations if posed that particular question. Naturally, we do not know how many patients, despite being posed that question, chose not to reveal symptoms of CBS.
We also performed a detailed examination of the retina to see whether there was a specific ocular pathology which could be associated with CBS. Abbott and Khan found no difference in the relative proportion of patients with neovascular AMD and geographic atrophy (Khan et al. 2008; Abbott et al. 2007). Abbott and co-workers also looked at relative sizes of central scotomas being unable to find a relation between the size and localization of scotomas and CBS (Abbott et al. 2007). All of our patients had neovascular AMD, and even though the size of the lesion was comparable in the two groups, we found that the patients with CBS had almost seven times larger areas of geographic atrophy than the other patients. This finding could suggest that larger areas of geographic atrophy from where there are no visual inputs to the visual cortex are more prone to cause hallucinations (Sunness et al. 2004). We did not find a correlation between area size of geographic atrophy and visual acuity – explained by the fact that some of the areas with geographic atrophy were extrafoveal. In contrast, we found thickness of subfoveal fibrosis correlated with visual acuity but not with the occurrence of CBS.
The origin of the CBS symptoms is still largely unknown even though the knowledge of the syndrome has recently been extended (Ffytche 2009). The main theory, denoted ‘deafferentation theory’, behind the syndrome suggests that a lack of sensory input from the eye creates hallucinations (Ffytche 2009). One case study linked anti-VEGF treatment with the onset of visual hallucinations and proposed that the hallucinatory episodes may be promoted by the reduced retinal oedema and realignment of the photoreceptors (Meyer et al. 2007). CBS can occur in any ophthalmological disease and has not been specifically associated with loss of function in the retina, lens or optic nerve. We feel that our data largely support this theory, suggesting that the hallucinations occur because of loss of visual input as may occur in geographic atrophy.
In conclusion, we show that symptoms of CBS are not uncommon in a large unselected group of patients being treated with anti-VEGF, and that a large proportion of the patients are reluctant to reveal their hallucinations unless they are specifically questioned about them. We also show that symptoms of CBS are associated with low vision and large areas of geographic atrophy. We feel that clinicians working with these patients should be aware of the relatively high prevalence of CBS in this population, so that the origin and benign nature of the hallucinations can be described to the patient.