Psychosis and Dementia—Disorders of Disadvantage

Abstract

As any Old Age Psychiatrist will tell you, late-life affective and psychotic disorders are some of the least understood mental health conditions, particularly considering their impact and the large number of people affected. This lack of attention partly lies in the common mundane situation of overshadowing research interests. Mental health research in older adults focuses almost entirely on dementia, while research into affective and psychotic disorders focuses almost entirely on young adults. Unfortunately, lack of research translates into an impoverished evidence base, which in turn filters through to underrepresentation in healthcare policy, and this compounds the problem by reducing the likelihood of interest from research funders, impeding capacity building and perpetuating a vicious cycle of inattention which does no favours for patients or health services. Even within the field, there are inequalities in research evidence, with considerably more attention paid to late-onset syndromes than to those who may have been living with their earlier-onset condition for decades by the time they reach older age ranges.

However, considering late-life psychosis, overshadowing research priorities are only part of the problem, as the conditions themselves present sizeable intrinsic challenges. The social withdrawal that characterises late-onset psychosis limits the likelihood of people affected being represented in conventional recruited research samples or retained in any study involving follow-up [1]. Furthermore, people with earlier-onset diagnoses are frequently out of contact with mental health services by the time they reach ‘older adult’ age ranges, so that it is difficult to identify potential participants for approaching in the first place. Conventional case control, cohort and intervention studies are therefore logistically formidable, if not wholly impractical and/or fundamentally limited by participation and/or attrition bias.

The growing availability of routine healthcare data for research provides important opportunities for improving the evidence base. Most national data governance frameworks permit the use of effectively anonymised healthcare information to be used for research which enables the better representation of populations who would previously have been considered ‘hard to reach’ in conventional recruited samples. This advantage is exemplified in the approach taken by Normark et al. [2], described in this issue, using national data from Denmark to investigate dementia incidence in people with schizophrenia and related disorders (ICD-10 F2x) according to their age at first psychosis diagnosis or treatment. There are reasonable considerations about the required reliance on recorded diagnoses (given that both conditions may be under-diagnosed, particularly dementia, and that under-diagnosis is unlikely to be at random); however, it is difficult to see how the question could be addressed in any other way.

Opinion on the relationship between late-onset psychotic disorders and dementia has changed over time, from early conclusions that there was no organic component in very-late-onset schizophrenia-like psychosis [3], to a growing realisation that it may be a frequent prodromal symptom of dementia, possibly similar to the more extensively described relationships between late-onset depression and dementia [4]. In particular, several studies have now confirmed that people with late-life psychotic disorder diagnoses have a higher incidence of dementia, with the risk often greatest in the first year after the onset of late-life psychosis [5, 6]. Furthermore, neuropathological change appears to be more common in late-onset schizophrenia suggesting that at least a subset of these individuals may be in the prodromal stages of neurodegenerative disease [7]. Indeed, early psychotic symptoms are associated with increased risk of developing Alzheimer's disease, in addition to a more rapid rate of cognitive decline [8]. However, less in known about the risk of dementia in people with psychotic disorders first diagnosed earlier in life, the objective for Normark et al. [2].

Normark et al. [2] also found a higher risk of dementia in people with earlier onset psychotic disorders diagnosed, confirming conclusions from previous meta-analyses [9]. This is not at all surprising. A multitude of adverse health outcomes have been described in people with psychotic disorders or ‘severe mental illness’, including long-recognised increased mortality generally and higher cardiovascular endpoints in particular [10]. These in turn reflect worse vascular risk factor profiles from diabetes, obesity and metabolic syndrome and adverse lifestyle factors such as physical inactivity, suboptimal diet and smoking. Increased dementia risk has been linked to all these factors, as well as to socioeconomic disadvantage more generally, and to social isolation, to which people with schizophrenia may be particularly prone. Early onset psychotic disorders are also associated with lower levels of education which contributes to reduced cognitive reserve and may, in turn, lower the threshold for dementia [11]. Risk may well be exacerbated further by anticholinergic and other (e.g., sedative) effects associated with antipsychotic use, and pre-dementia reserve may also be reduced as a consequence of the cognitive impairment associated with at least some psychotic disorders including deficits in attention, working memory and executive function.

Neurodegenerative change is another factor which may increase the risk of disorders underlying dementia in a variety of ways, including vascular dementia from small and large cerebral infarctions, as well as Alzheimer's disease induced by pro-inflammatory states and/or accelerated in onset because of more minor levels of cerebrovascular disease and reduced cognitive reserve. However, a recent meta-analysis concluded similar rates of AD pathology in individuals with cognitive impairment associated with chronic psychosis relative to healthy controls [12], suggesting that the dementia associated with early-onset psychosis is not primarily driven by neurodegenerative change. It seems likely that the mechanisms driving the increased dementia risk in early- and late-onset psychosis may differ, with neurodegenerative pathological change a less salient factor in individuals who develop psychosis earlier in life.

It could be said that the contribution of epidemiological research over the last 20–30 years has been to reconceptualise both psychosis and dementia as ‘disorders of disadvantage’. For dementia, after early paradigms restricted to genetics and discrete risk factors like head injury, evidence from the mid-1990s broadened understanding of aetiology to include vascular disorders and adverse lifestyles, all of which have long been recognised to be strongly socially determined and to accumulate from birth onwards, alongside education and early-life cognitive attainment determining later brain reserve. Indeed, there are few common risk factors for premature mortality that are not also found to be associated with higher dementia risk in those who survive, so the association with psychosis found here is simply fitting in with the broader picture. For psychosis, alongside a growing appreciation of the roles of social factors in its own development, a previously narrow focus on mental health outcomes has similarly expanded over the last few decades to encompass the much wider health disparities faced by those affected. The association with increased risk of dementia observed here, is therefore again simply fitting in as a final consequence of prolonged disadvantage.

So, where does this take us? Recent consensus statements have emphasised the potential to prevent many cases of dementia [13], if only by delaying its onset so that it does not occur (or occurs in only mild stages) in life. And the required actions for prevention are principally those that would be recommended for any health improvement—better diet, physical activity, social support, well-controlled health conditions, and so forth. People with psychosis stand to benefit most from these health improvement initiatives but are frequently left with least access. Whether adding dementia prevention to the list of potential benefits would help improve access is uncertain but seems at least worth evaluating. More immediately, the higher baseline risk of dementia is worth bearing in mind in clinical assessments of older people with established schizophrenia and other psychotic disorders, adopting a higher level of caution when cognitive impairment is present and being careful not to assume that it is intrinsic to the condition unless there are strong reasons to support this. For late-onset psychosis, as for late-onset depression, the possibility of a pre-dementia prodrome needs to be considered particularly seriously. Psychiatric prodromal symptoms are recognised for dementia with Lewy bodies [14]; however, despite their frequency, they are poorly conceptualised for other dementia subtypes (and are not adequately captured by the ‘mild behavioural impairment’ concept). It is noteworthy that dementia hazards observed by Normark et al. [2] were strongest in those with older age at psychosis diagnosis and younger age at dementia diagnosis, supporting a strong role of psychosis as a prodromal symptom in at least some individuals.

Finally, there are clearly many more questions to answer. While the relationship between psychosis and risk of dementia can be ascertained from information conventionally contained in large national healthcare databases, the next set of questions will require more granular data resources. For example, understanding the potential role of different vascular risk factors requires linked data that capture these over considerable time periods, bearing in mind the changes that occur in factors such as weight, blood pressure and cholesterol levels during the early stages of dementia development [15]. Characterising the aetiology of any incident dementia is important, and the increasing use of biomarkers to support diagnosis and confirm evidence of any neuropathological changes will be key to answering this question. Furthermore, understanding the clinical transition from psychosis to dementia may well require source information on measured cognitive function, symptom profiles and social support, that is typically represented in text rather than pre-structured fields in source records. However, advances are being made in opening up this level of data granularity [16], and new developments in natural language processing applications will allow better clinical characterisation of phenotypes and longitudinal changes in large datasets, providing at least some chance of better representation for late-life functional mental disorders. Future progress will depend on harnessing these innovations and enabling a clearer understanding of the diverse pathways linking psychosis and dementia. Understanding these pathways is not only essential for scientific clarity but also to deliver equitable care and dementia prevention to people with early- and late-onset psychosis.

The authors declare no conflicts of interest.