Author
BRUNELLE Sarah; COLE Martin G.; ELIE Michel;
Title
Risk factors for the late-onset psychoses: a systematic review of cohort studies.
Journal citation/publication details
International Journal of Geriatric Psychiatry, 27(3), March 2012, pp.240-252.
Summary
Eleven studies, mostly conducted in Europe, examined a total of 32 potential risk factors. Although some evidence of an association with the development of late-onset psychosis was available for a few risk factors, the main conclusion was that the lack of availability of methodologically sound literature limits the validity of the findings.
Context
As many as a quarter of patients who develop schizophrenia do so at the age of 40 or later. The risk factors for the onset of schizophrenia in younger patients are relatively well established but little research has been published on risk factors for late onset psychoses, using evidence from cohort studies.
Methods
What sources were searched?
Three electronic databases, EMBASE plus EMBASE CLASSIC, Medline, and PsycINFO, were searched from the date of inception to March 2010. Additional articles were sought by screening the reference lists of potentially relevant articles and contacting experts in the field.
What search terms/strategies were used?
Searches were carried out using a combination of the keywords ‘late-onset psychosis’ and ‘risk factor*’ and applying the age limits ‘40-60 years’ and ‘65 years and over’.
What criteria were used to decide which studies to include?
Primary research using a longitudinal prospective, or retrospective, cohort and reporting at least one risk factor or predisposing factor, was eligible for inclusion. The population studied had to include people with a mean age of 40 years, or older, at the onset of psychosis, or a subgroup analysis for the age group. In addition, an outcome of a psychotic disorder or symptoms from the following list was required: schizophrenia, paraphrenia, delusional disorder, schizophreniform disorder, schizoaffective disorder, psychotic disorder not otherwise specified, paranoid psychosis, delusions, paranoid ideations or hallucinations. Studies were excluded if at the time of diagnosis one of the following was also present: an active mood disorder, dementia, delirium, or a mini mental state score below 24 out of 30. Articles had to be available in English or French.
Who decided on their relevance and quality?
The full text of potentially relevant articles, identified by screening titles and abstracts, was examined against the selection criteria by the first author. The process was independently checked by the third author. Study quality was assessed by at least two of the authors working independently, using a five-point scale developed by Gray, 2004. Differences were resolved by discussion. Most of the studies met at least three quality criteria, but none met all five.
How many studies were included and where were they from?
Eleven studies published in 12 publications were selected for inclusion from 5,887 articles retrieved from the database searches, 141 reference list papers and two articles contributed by experts. All of the studies, apart from two set in the USA and one from Australia, were conducted in Europe.
How were the study findings combined?
Data abstraction and synthesis was carried out by at least one co-author blind to previous results. All reported risk factors were organised into common categories and incorporated into a data abstraction checklist. Results were abstracted as odds ratios of relative risk, with confidence intervals, or as narrative data. A risk factor was considered to be positive if one item in the category was significant for at least one of the psychotic outcomes listed as an inclusion criterion. Results are reported separately for univariate and multivariate analyses. Pooled estimates for the odds of late-onset psychosis associated with individual risk factors were calculated using a random effects model.
Findings of the review
Study population size ranged from 93 to over one-and-a-half million; the proportion of females included in the study samples varied from 44.6% to 77.3%. Psychotic symptoms and disorders were measured using a wide variety of criteria, scales and structured interviews. In total, 27 potential risk factors were reported on univariate analysis and 32 were identified by multivariate analysis. The most frequently cited were increasing age and female gender, reported in nine studies each.
Significant risk factors, identified in at least 60% of two or more studies on multivariate analysis, were - having a positive history of psychosis, cognitive problems, poor physical health, and visual impairment. There was also evidence that negative life events, especially in studies reporting a higher number of events and events of severe intensity, have a negative impact on the development of psychotic symptoms.
Factors previously reported in cross-sectional studies to have an association with a greater risk of late-onset psychosis that were not confirmed in this review were older age, female gender and hearing problems. The role of social isolation as a risk factor was also not confirmed. Only data on one potential risk factor, female gender, was subjected to meta-analysis. The estimated combined odds ratio, based on data from six studies, was 1.160 (95% confidence interval -1.673, p -0.429).
Authors' conclusions
‘Research on the late-onset psychoses is very heterogeneous and few longitudinal studies are available. Previously postulated risk factors for late-onset psychosis such as older age, female gender, and hearing deficit did not appear significant in cohort studies. The strongest predictors of psychosis in late adulthood were past psychotic symptoms, visual impairment, cognitive deficit, poor health status, and negative life events.’
Implications for policy or practice
None are discussed.