Objective: Visual hallucinations (VH) most commonly occur in eye disease (ED), Parkinson's disease (PD), and Lewy body dementia (LBD). The phenomenology of VH is likely to carry important information about the brain areas within the visual system generating them. Methods: Data from five controlled cross-sectional VH studies (164 controls, 135 ED, 156 PD, 79 (PDD 48 + DLB 31) LBD) were combined and analysed. The prevalence, phenomenology, frequency, duration, and contents of VH were compared across diseases and gender. Results: Simple VH were most common in ED patients (ED 65%
METHODS
Study Selection and Data CollectionThere have been several studies(2-5, 8, 24) using NEVHI(3) since its original publication in 2008. Only controlled cross-sectional NEVHI studies with ED, PD, PDD, and DLB samples using similar methodology were included in this study. Three(3, 4, 8) of them had two sample groups (patients, controls), while the other two had three sample groups (controls, ED, DLB;(2) controls, PD, PDD (5) Committee, United Kingdom. All procedures related to the study were explained to the participants and a written informed consent was obtained prior to participation. All data were merged into a single database including 164 individuals in the comparison group and 383 patients (135 ED, 156 PD, 48 PDD, 31 DLB, 13 with combination of ED/PD/PDD/DLB). The latter 13 patients were excluded from further analysis. A sub-analysis of VH phenomenology did not reveal any differences between DLB and PDD patients, which were subsequently considered as one LBD group. PD and LBD patients were only included if they had no visual field defects on neurological examination.
AssessmentsAn interview was used to gather demographic data in all studies. Best near visual acuity was examined 6 members/children), "body parts", "animals/insects", "machines", and "letters/numbers/musical notes".(7, 13, 15) Prevalence, phenomenology, frequency, duration and severity of VH were compared between the (i) comparison group and patients and (ii) different patient groups (ED, PD, LBD). The phenomenology of Complex VH was also compared between gender.
Statistical analysisStatistical analysis was undertaken using the Statistical Package for Social Sciences (SPSS Version 20).Normal distribution of data was examined using the Shapiro-Wilk test. Means and standard deviations (SD) were calculated. Data were analysed using parametric tests (one-way analysis of variance (ANOVA)). Multiple comparisons were assessed with Post-hoc Bonferroni tests. Frequencies were compared using the chi-square (χ 2 ) test and Fisher's exact test when expected frequency in either group was < 5. The effect of gender, disease and interaction of gender × disease on contents of complex VH was tested using a two-way ANOVA (Effect sizes are reported using partial η 2 ), whilst logistic regression was used to ascertain whether the VH phenomenology could significantly predict the disease. The homogeneity of variance of the interaction model was tested using Levene's test, while go...