Results

The ApoE genotyping was ε2/3 in eight patients, ε2/4 in three patients, ε3/3 in 78 patients, ε3/4 in 115 patients, and ε4/4 in 24 patients. Thus, the ApoE ε4 gene was positive in 142 patients and negative in 86 patients. As expected, the age at examination was highly correlated with the age at onset (r=0.96, p < 0.0001). The educational attainment was inversely correlated with both age at examination (r=−0.31, p<0.0001) and age at onset (r=−0.32, p<0.0001). Both mean age at examination and age at onset were significantly greater in women than in men (73.3 (SD 7.8) v 70.7 (9.2), p=0.046; 70.4 (8.30) v67.5 (9.37), p=0.034)). The mean educational attainment was significantly less in women than in men (8.6 (SD 2.1) v9.6 (2.3), p=0.0048)). Although age, education level, and the MMSE score did not differ significantly between the ApoE ε4 positive patients and the ε4 negative patients, the mean duration of illness was significantly greater in the ε4 positive patients than in the ε4 negative patients (39.0 (SD 30.9) v 30.9 (23.6) months, p=0.038).

Of the 228 patients, 118 (51.8%) were rated as psychotic. Of these 118 patients, 21 had both delusions and hallucinations, 94 had delusions only, and three had hallucinations only. Thus 115 (50.4%) of the patients had delusions. Of these, 91 patients had persecutory delusions (88 patients had delusions that “people are stealing things”, 10 patients had delusions of “being conspired against or harassed”, and three patients had a delusion of abandonment), 58 patients had misidentification delusions (42 patients had delusions that “someone was in the house”, 11 patients had delusions that “the house was not their own”, six patients had delusions that “television figures are actually present in the home”, and six patients had other types of misidentification), three patients had delusional jealousy, one patient had a delusion of poverty, one patient had grandiose delusions, and one patient had eromanic delusions. Some patients had two or more kinds of delusions. Twenty four (10.5%) of the patients had hallucinations. Of these, 16 had visual hallucinations, 14 had auditory hallucinations, and one had olfactory hallucinations. Seven patients had two types of hallucinations.

Table 1 compares the characteristics of patients with and without psychosis. Those with the psychosis were significantly older, had an older age of onset, had a longer duration of illness, and had less education than those without psychosis. The proportion of women was significantly higher in the psychotic group than in the non-psychotic group. There were no significant associations between psychosis and ApoE genotype. The mean MMSE score was significantly lower in patients with psychosis than in the those without psychosis.

Table 2 shows the results of the first model of multiple logistic regression in which age at onset was included, and table 3 shows that of the second model including age at examination. Both models indicated that the effects of age, sex, and MMSE were significantly associated with the presence of psychosis. In the first model but not in the second, duration of illness was a significant risk factor for psychosis. The level of education was not a significant predictor of psychosis in either model. The ApoE genotype also had no significant association with psychosis.

Discussion

The present study provides further evidence that psychotic symptoms are a common manifestation of Alzheimer’s disease. The prevalences of delusions (51.8%) and hallucinations (10.5%) were well within the ranges reported in the literature.1 2 The finding that persecutory delusions (especially the belief that people are stealing or hiding objects) were the most common psychotic manifestations was also compatible with the findings of previous studies.9 12 14 16 17 20 33 Because patients who had parkinsonism were excluded from this study, the chance of erroneous inclusion of those with dementia with Lewy bodies,34 in whom neuropsychiatric symptoms are common, was small. The major findings of our study were that sex, age, and severity of cognitive impairment were each significantly and independently associated with psychosis, and that the level of education and the frequency of the ApoE ε4 allele were not significant predictors of the presence of psychosis.

Previous studies have disagreed on the contribution of cognitive impairment in psychotic symptoms. Some studies failed to show a difference in cognitive impairment between the patients with and without psychosis,7 10 11 14 15 35-37 whereas others showed that psychotic symptoms were associated with poor13 17-19 21 or preserved16 20cognitive function. These controversies are likely due to the inclusion of patients with different severities of dementia. Psychotic symptoms were considered to be uncommon in both the very mild33 and very severe18 20 stages of Alzheimer’s disease. A longitudinal study2 disclosed that delusions in patients with Alzheimer’s disease were relatively more frequent during the moderate stage. The small proportion of advanced patients included in our study contributed to the positive correlation between psychosis and cognitive impairments. In any case, as the difference in MMSE between groups was small, the effect of cognitive impairment in psychosis would be modest. Nevertheless, it is conceivable that a dysfunction of specific cognitive domains is associated with psychosis.

In this study, duration of illness was shown to be an independent predictor of psychosis in the logistic regression model in which age at onset was the variable, although it was not significant in the model when age at examination was the variable. It is plausible that the effect of duration of illness was masked by the effect of the age at examination, because the age at examination is a function of age at onset and duration of illness. Our results supported the conclusion of the study of Migliorelli et al,15 who reported that patients with Alzheimer’s disease with delusions had a significantly longer duration of illness than those without delusions, whereas most previous studies failed to show a significant association between duration of illness and psychosis.8 10 13 18 20 21 35-38 An association between development of psychotic symptoms and a rapid mental deterioration has been found in several longitudinal studies.2 9 33 39 40 This does not disagree with our finding that psychotic patients with Alzheimer’s disease had a significantly longer duration of illness, because long duration does not necessarily imply slow progression of the disease. Our findings that a decreased MMSE and a longer duration of illness were separate factors associated with the presence of psychosis would suggest that deterioration of both cognitive and non-cognitive functions promotes psychosis independently.

Older age was shown to be a significant risk factor for psychosis in both logistic regression models in this study. The issue of the effect of age on psychosis in Alzheimer’s disease is highly controversial. Several investigators7-11 showed that older patients were more likely to have psychotic symptoms, whereas Gilly et al 8 reported that early onset was significantly associated with psychosis when cognitive functions were severely impaired. Devanand et al 13 maintained that age at examination was associated with paranoid delusions and inversely correlated with hallucinations. Moreover, many studies failed to show a significant relation between age and psychosis.12 15 17-21 35-38 41 42 However, evidence that supports the relation between older age and development of psychosis is available in studies focusing on localisation of pathological changes in relation to age and psychosis. Senile plaques both in the temporal neocortex and in the hippocampus were more frequent in younger patients with Alzheimer’s disease than in elder patients,^43 44 whereas psychosis in Alzheimer’s disease is reportedly inversely related to the preservation in those areas.45

We also found a significant association between female sex and development of psychosis. Although female sex has been implicated as a risk factor for Alzheimer’s disease, few studies have considered whether sex specifically predisposes patients to the development of psychotic symptoms in Alzheimer’s disease. Devanand et al 13 reported that women’s scores of delusion on the Columbia University scale for psychopathology were significantly higher than those of men. Reisberg et al 12found that women tended to exhibit behavioural symptoms more often. On the other hand, Burns et al 14 reported a higher prevalence in men of delusions of theft. Other studies failed to show an association of psychosis with sex.10 11 15 20 21 35 37 38 41 46 However, in other domains of neurological disease, there is evidence of female sex as a risk factor of psychosis. Comparing psychotic and non-psychotic epileptic patients, Taylor47 showed that female sex was a risk factor for development of psychosis. Our finding and those of others remain unexplained, and relevant factors involving genetic traits, as well as cultural and psychosocial involvements, should be investigated.

The multivariate analysis showed that the level of education was not an independent protection factor for psychosis. The association between psychosis and fewer years of education shown in the univariate analysis is likely to be an epiphenomenon which can be accounted for by the inverse correlation between the educational attainment and age and by the negative association between the level of education and female sex. Although one study13 reported that education was inversely related to psychosis, the results of our multivariate analysis agreed with most studies, which denied this effect.10 11 15 18 20 21 33 41

It has been postulated that patients with Alzheimer’s disease with and without the ε4 allele have distinguishing phenotypic characteristics.42 Ramachandran et al,22 in a preliminary report, showed that patients with Alzheimer’s disease and the ApoE ε3/4 genotype had more than a threefold increase in psychosis when compared with those with the ApoE ε3/3 genotype. As female patients with late onset Alzheimer’s disease carried the ε4 allele more often than did male patients,31 their finding might be attributable to the effects of sex and age. Lehtovirta et al,42Lopez-Alberola et al,48 and Lopez et al 49 showed no association between the ApoE ε4 gene and development of psychosis. In our study, in which other variables were concurrently examined, this association was clearly denied.

The present study showed that a subpopulation of patients with Alzheimer’s disease who are female, of late onset, of severe cognitive impairment, and of long duration of illness, have a high risk for development of psychosis. One possible explanation for our findings is that a particular subtype of Alzheimer’s disease which preferentially involves old women has distinctive clinical and neuropathological features. An alternative explanation is that characteristics including genetic, psychosocial, and environmental factors may affect development of psychosis in individuals. Further studies are needed to consider these hypotheses.