Poststroke psychosis: a systematic review

A preregistered systematic review of poststroke psychosis examining clinical characteristics, prevalence, diagnostic procedures, lesion location, treatments, risk factors and outcome. Neuropsychiatric outcomes following stroke are common and severely impact quality of life. No previous reviews have focused on poststroke psychosis despite clear clinical need. CINAHL, MEDLINE and PsychINFO were searched for studies on poststroke psychosis published between 1975 and 2016. Reviewers independently selected studies for inclusion, extracted data and rated study quality. Out of 2442 references, 76 met inclusion criteria. Average age for poststroke psychosis was 66.6 years with slightly more males than females affected. Delayed onset was common. Neurological presentation was typical for stroke, but a significant minority had otherwise ‘silent strokes’. The most common psychosis was delusional disorder, followed by schizophrenia-like psychosis and mood disorder with psychotic features. Estimated delusion prevalence was 4.67% (95% CI 2.30% to 7.79%) and hallucinations 5.05% (95% CI 1.84% to 9.65%). Twelve-year incidence was 6.7%. No systematic treatment studies were found. Case studies frequently report symptom remission after antipsychotics, but serious concerns about under-representation of poor outcome remain. Lesions were typically right hemisphere, particularly frontal, temporal and parietal regions, and the right caudate nucleus. In general, poststroke psychosis was associated with poor functional outcomes and high mortality. Poor methodological quality of studies was a significant limitation. Psychosis considerably adds to illness burden of stroke. Delayed onset suggests a window for early intervention. Studies on the safety and efficacy of antipsychotics in this population are urgently needed.


Databases and search strategy
CINAHL, MEDLINE and PsychINFO were electronically searched from January 1975 to December 2016. We judged the 1st of January 1975 to be a natural starting date for the search, as studies published prior to this date are unlikely to involve neuroimaging techniques that provide critical information for this type of research, i.e. CT scans. The final search was completed on 9 December 2016. The search strategy was limited to only include studies published in English and studies assessing human participants.

Data extraction and quality assessment
Two reviewers independently screened the titles and abstracts of the literature identified by the search strategy described above, excluding studies that did not meet the inclusion criteria.
During this process any study whose eligibility could not be determined after having reviewed the title and abstract was obtained via full text format, and included in the next stage of the search process. The full text of the remaining studies were then screened by the same reviewers and again excluded if they did not meet the inclusion criteria. Any uncertainty or disagreement about the studies at this stage was discussed between the two reviewers in order to reach consensus, or with the third author.
Finally, two reviewers used a standardised extraction form to extract relevant data from the included studies. Data was extracted with respect to study design, patient setting and patient characteristics, including age and number of participants, and whether or not the participants had been diagnosed with a psychiatric disorder prior to having a stroke. Moreover, information regarding each study's inclusion and exclusion criteria, comparison group or condition, and diagnostic assessment was extracted. Specifically, information about stroke subtype, lesion location, as well as time of onset and description of psychosis was obtained. If a study explored treatment interventions the method for assessing treatment outcomes was also extracted, as was each study's primary outcome and other outcomes. No authors were contacted for additional information.
Quality assessment was conducted using a revised version of the STROBE checklist for evaluating observational studies. Ten items were adapted from the STROBE checklist as there is currently no consensus about what assessment tool is appropriate to use for qualitative studies.
Each item was scored either "Yes" or "No" by two reviewers and a global score was calculated for each item. The quality assessment is available below.

Reasons for exclusion
After having screened the titles and abstracts of the studies identified by the search strategy, 198 were identified as potentially eligible. Out of these, 122 studies were excluded after having been obtained the full text, as they did not meet the inclusion criteria for the review. 41 of these studies were excluded because they did not meet the inclusion criteria for psychosis.
Typically these studies involved non-psychotic hallucinatory symptoms due to sensory impairments, such as Charles Bonnet syndrome, or delusions relating to body ownership or body awareness, such as somatoparaphrenia and anasognosia. A total of 35 studies were excluded because the causation between stroke and psychosis was unclear, for example where patients were presenting with a comorbid neurodegenerative disease. A further 18 studies were excluded because they did not involve psychosis, and 13 studies were excluded due to unclear psychosis definitions, such as symptoms presented being representative of affective disorders rather than psychosis. We excluded 6 studies because they did not involve stroke pathology and 5 studies because they did not include CT or MRI scans to verify a diagnosis of stroke. Finally, 3 studies were excluded because they were reviews and did not report any original data, and 1 study was a personal account.

Meta-analysis of prevalence
Three studies reported prevalence of delusions and hallucinations in stroke patients. Kumral and Öztürk (2004) reported on delusions in a sample of acute stroke patients, Buijck et al (2011) reported prevalence rates for delusions and hallucination on admission and discharge in a sample of geriatric patients admitted for stroke rehabilitation, and van Almenkerk et al (2012) reported prevalence rates of delusions and hallucinations in a sample of institutionalized stroke patients.
To estimate cross-sectional prevalence we only included prevalence at the point of admission Because of the methodological differences in Kumral and Öztürk (2004) and concerns about adequate screening for patients with alterations in their levels of consciousness, not least because they various report patients with agitation and two specifically with confusional states, we also report meta-analytic results with this study removed, using the same methods, in Figure S1 below, leading to very similar estimates of prevalence. Figure S1. Forest plot of post-stroke delusion and hallucination prevalence with Kumral and Öztürk (2004) removed

Quality assessment of the case series and quantitative studies
All of the 18 studies included an adequate description of the setting and location of the studies, and specified the eligibility criteria and sources and methods for the selection of the participants. A total of 13 (72.2%) studies included definitions of psychosis as measured by diagnostic criteria, and 6 (33.3%) specified the diagnostic criteria used to diagnose stroke. In 11 (61.1%) of the studies stroke data were obtained from a patient or proxy, whereas in 7 (38.9%) stroke data was collected from a database. A total of 11 (61.1%) studies clearly defined their neuroimaging findings, and 3 (75%) of the eligible studies included a blinded investigator. We found that 17 (94.4%) of the studies adequately reported the number of participants at each stage of the study. All of the studies reported appropriate descriptive and outcome data. Finally, 6 (50%) of the eligible studies provided unadjusted estimates and their precision (e.g. 95% confidence interval) in their main results. Full results are displayed in Table 3.  (1996) 19/M I Schizophrenia-like n/a n/a Risperidone Partial resolution

Participants
Report numbers of individuals at each stage of study-eg numbers potentially eligible, examined for eligibility, confirmed eligible, included in the study, completing follow-up, and analysed Descriptive data Give characteristics of study participants (eg demographic, clinical, social) and information on exposures and potential confounders.
Outcome data Report numbers of outcome events or summary measures.

Main results
Give unadjusted estimates and, if applicable, confounder-adjusted estimates and their precision (eg, 95% confidence interval). Make clear which confounders were adjusted for and why they were included.