Clinical presentation

Previous literature has suggested poststroke psychosis is ‘rare’,4 although the meta-analytic summary presented here suggests that either delusions or hallucinations with poor insight occur in an estimated 4.86% of patients with postacute stroke. Prevalence is therefore likely greater than previously assumed, although we note the need for additional studies in this area, as the estimate reported here was based on a total of four studies, only three of which used a structured assessment of symptoms.

The average time between a stroke and the onset of psychosis or psychotic symptoms was reported in two studies, although with differing results. Almeida and Xiao8 reported an average time to onset for psychosis of 6.1 months over a follow-up period of 10 years. Focusing specifically on delusions after acute admission, Kumral and Oztürk9 reported an average time to onset of 2 days with a short duration but with presentations also including agitation, aggression, emotionalism, anxiety reactions and depression. Although only two patients are reported as having acute disorientation, given the other presenting features, acute admission population and use of clinical diagnosis, it is not clear to what extent they successfully screened out patients with fluctuating consciousness or confusional states when assessing for delusions, and we suspect Almeida and Xiao8 provide a better estimate of psychosis onset.

It is worth noting that studies on traumatic brain injury (TBI) have also reported a significant delay from point of insult to psychosis onset, with psychosis becoming more likely 2–3 years the index injury.22 This is additional evidence that although somatoparaphrenia and anosognosia are often classified as delusions, they appear aetiologically distinct from poststroke psychosis because they typically occur in the acute poststroke stage and quickly resolve.23 In contrast, poststroke psychosis (like post-TBI psychosis) seems often to occur after a significant delay, suggesting that each could be understood as arising from disruption to distinct stages of the recovery trajectory after brain insult.

A total of 18.1% of the patients in this review presented with haemorrhagic strokes, whereas 79.8% presented with ischaemic strokes. These findings are comparable with those found in the general stroke population, where haemorrhagic strokes normally account for approximately 20% of strokes and ischaemic strokes account for the largest proportion24 suggesting that stroke type is not a predictor of subsequent psychosis. The average age for the occurrence of poststroke psychosis was 66.6 years, and it occurred more often in men than in women. In terms of gender balance, these findings are in contrast to the wider stroke population, where women show a higher stroke prevalence.25 It is not clear why men might show a higher rate of poststroke psychosis, and this prevalence does not track the background risk factors for psychosis where men have a higher lifetime risk of developing primary psychotic disorders, such as schizophrenia,26 but women are over-represented in cases of late onset psychosis, which occur after the age of 60 years.27 Compared with poststroke depression, which has been more extensively researched, subsequent depression is more common in women. However, the interaction between gender and stroke risk is complex and involves historical, neurological and social risk factors.28 We expect a similar complex picture to occur in poststroke psychosis as more systematic evidence accumulates.

The majority of patients were found to have right hemisphere lesions, in line with a large body of evidence that suggests right hemisphere pathology is associated with a range of perceptual anomalies and pathologies of belief.29 The regions that were most affected included the right frontal, temporal and parietal regions, as well as the right caudate nucleus, supporting findings of previous case series on poststroke psychosis10–12 and mirroring findings in psychosis after TBI.21 However, it is worth bearing in mind that left hemisphere strokes are more likely to result in language and communication problems, which may impede the assessment of psychosis, and therefore these patients may be under-represented in the literature.30 It is also worth noting that these findings are largely based on clinical radiological assessments. To date, we are only aware of two studies that have used voxel-based lesion mapping across patients with stroke to generate maps of common lesions associated with psychosis. Devine et al 11 examined lesion location in three patients with poststroke psychosis finding that damage to the right inferior frontal gyrus and connecting white matter tracts were common across patients. More recently, Darby et al 31 used lesion network mapping in 17 stroke patients with misidentification delusions, identifying the left retrosplenial cortex as the only brain area functionally connected to all lesion sites, potentially highlighting its role in the aetiology of poststroke psychosis. Additional studies using these more systematic approaches are clearly needed to better clarify how damage to specific areas and neural circuits could lead to psychosis.

The most common presenting neurological signs and symptoms accompanying poststroke psychosis were left-sided weakness, headache and slurred speech. Others included left-sided neglect, left-sided decreased vision or blindness, left-sided facial weakness and confusion. Notably, these are all common presenting symptoms of stroke, and it is unclear to what extent they can be considered to be specific predictors of poststroke psychosis beyond indicating right hemisphere lesion.32 Importantly, however, a significant minority of patients were found to have otherwise ‘silent strokes’, where neurological signs were not obvious, not present or were subtle enough to be missed during referral to a psychiatrist for psychosis, suggesting that stroke may go undetected for a considerable amount of time. It is unclear to what extent the development of psychosis might affect the later chances of stroke detection, given that psychosis is typically associated with a significant delay in accessing treatment.33

The most common types of psychosis were, in order of frequency, delusional disorder, schizophrenia-like disorder and mood disorders with psychotic features. Notably, delusional disorder was more common than schizophrenia-like psychosis in this population, despite the fact that ‘non-organic’ delusional disorder has been found to have a relatively low lifetime prevalence in the general population.34 35 Evidence that delusional disorder and schizophrenia-like disorder are also the most common types of psychosis to follow after TBI36 suggests that more circumscribed lesions may produce less systematised, monothematic delusions that are less likely to be accompanied by hallucinations.

In terms of delusional theme, persecutory delusions were the most common, followed by delusional jealousy (Othello syndrome), reduplicative paramnesia and somatic delusions. Delusional jealousy seems to be relatively common here: although persecutory delusions and reduplicative paramnesia are also common after TBI, delusional jealousy is not as frequently reported.36 However, in comparison with its prevalence in schizophrenia and bipolar disorder, delusional jealousy seems to be more common in neurological disorders and neurodegenerative diseases for reasons that are still unclear.37

The majority of patients did not have a previous psychiatric history, although this was somewhat inconsistently reported across studies. Other risk factors included hypertension, hyperlipidaemia and diabetes, which are consistent with general risk factors for stroke.38

Diagnostic categories for poststroke psychosis are used inconsistently in the literature, and it is not clear which rating scales are most appropriate in this population, introducing a barrier to more systematic research in the area. Indeed, only 11 of the included studies applied diagnostic criteria, as specified by the Diagnostic and Statistical Manual of Mental Disorders and International Classification of Diseases, and only 12 studies included information about what psychosis assessment measures were used. This lack of structured assessment makes it difficult to draw more precise conclusions concerning differences in the presentation and severity of psychosis. Our review finds that clinicians have few options when faced with people in the clinic, and there is a need for a more structured assessment of psychosis.

Treatment and outcome

No studies formally evaluated treatments for poststroke psychosis, but the available evidence suggests that antipsychotic medication is the most common form of treatment. Haloperidol and risperidone are reported as the most frequently prescribed, followed by quetiapine and olanzapine. It is worth noting that these types of treatments are linked to side-effect profiles that include dyskinesias and weight gain, and some, especially risperidone, are themselves a risk factor for developing stroke.39 It is therefore a cause for concern that so many individuals with poststroke psychosis are prescribed antipsychotic medication when no clinical trial has evaluated clinical effectiveness, safety or adverse effects. Our review identifies this as a research priority.

Psychological treatments were reported only in a few studies, indicating that psychological intervention as a treatment for poststroke psychosis remains largely unexplored. This is despite the fact that neuropsychological rehabilitation and psychological intervention are commonly used with people affected by stroke and psychosis, respectively, suggesting an opportunity for non-pharmacological management.

It is worth noting that the most common treatment outcome reported in the literature was complete resolution of poststroke psychosis, with the average time interval to complete resolution 3.5 months. These findings suggest that standard psychiatric management for poststroke psychosis is a reasonable approach. However, it remains unclear to what extent publication bias has played a role in presenting positive cases, how poststroke psychosis would resolve in the absence of treatment and what types of treatment are most effective.

Nevertheless, the evidence reviewed here suggests that long-term outcomes of poststroke psychosis are generally poor, indicating that patients with poststroke psychosis may have greater difficulties coping with the consequences of stroke and more likely to depend on assistance in their everyday lives compared with other stroke survivors. Furthermore, preliminary evidence suggests that patients with poststroke psychosis have an increased 10-year mortality risk.8 The reasons for this are currently unclear, and additional work is needed to understand to what extent this is due to the complicating effect of psychosis and to what extent due to the ongoing effects of background risk factors for psychosis itself. However, unhealthy lifestyle choices, antipsychotic side effects, suicide and physical illnesses, including cardiovascular illnesses, have all been found to contribute to excess mortality in primary psychosis,40 and we hypothesise that similar factors may affect those with poststroke psychosis.

Limitations

Despite our systematic approach to this review, several limitations should be noted in terms of the findings and wider evidence base. The single most important limitation is the methodological quality of studies reviewed, which was generally poor. The majority were single cases, with group studies typically reporting small sample sizes. Moreover, only nine studies included a control group, and only one of the included studies was a prospective cohort study. In addition, the poor quality of reporting (in case studies particularly) meant that many key details needed to complete an adequate quality assessment were missing.

Given this, it is important to highlight that numerical summaries of published cases may be a poor guide to actual characteristics in the stroke population. In one instance, namely stroke type prevalence, the frequencies in published cases match estimates from epidemiological studies quite closely. However, it is likely that summaries of published case studies will be poorly representative of many aspects of poststroke psychosis, particularly, we suspect, regarding the type of delusion or hallucination, where rarer presentation may be more commonly reported in the literature, and treatment outcome, where patients with poorer outcome are likely to be under-represented. Currently, there are no systematic treatment studies, and this remains a marked deficiency in the literature and a significant barrier to evidence-based care. With this in mind, any characteristics of poststroke psychosis described in this review that are not echoed by systematic studies need to be treated, at best, as preliminary, until these studies are completed.

It is also worth noting that the inclusion criteria used in this current review was specifically designed to include true cases of psychosis (delusions and/or hallucinations with limited or absent insight). However, this meant that patients with poststroke hallucinations with full insight into the nature of their hallucinatory experience (hallucinosis) were not included, and it is not clear to what extent this group differs from patients with poststroke psychosis.