Associations between multimorbidity and neuropathology in dementia: a case for considering functional cognitive disorders, psychiatric illness, and dementia mimics

Background Multimorbidity, the presence of two or more health conditions, has been identified as a possible risk factor for clinical dementia. It is unclear whether this is due to worsening brain health and underlying neuropathology, or other factors. In some cases, conditions may reflect the same disease process as dementia (e.g. Parkinson’s disease, vascular disease), in others, conditions may reflect a prodromal stage of dementia (e.g. depression, anxiety and psychosis). Aims We aimed to assess whether multimorbidity in later life was associated with more severe dementia-related neuropathology at autopsy. Method We examined antemortem and autopsy data from 767 brain tissue donors from the United Kingdom, identifying physical multimorbidity in later life, and specific brain-related conditions. We assessed associations between these purported risk factors and dementia-related neuropathological changes at autopsy (Alzheimer’s disease related neuropathology, Lewy pathology, cerebrovascular disease, and limbic-predominant age-related TDP43 encephalopathy) with logistic models. Results Physical multimorbidity was not associated with greater dementia-related neuropathological changes. In the presence of physical multimorbidity, clinical dementia was less likely to be associated with Alzheimer’s disease pathology. Conversely, conditions which may be clinical or prodromal manifestations of dementia-related neuropathology (Parkinson’s disease, cerebrovascular disease, depression, and other psychiatric conditions) were associated with dementia and neuropathological changes. Conclusions Physical multimorbidity alone is not associated with greater dementia-related neuropathological change; inappropriate inclusion of brain-related conditions in multimorbidity measures and misdiagnosis of neurodegenerative dementia may better explain increased rates of clinical dementia in multimorbidity. Data Set Information Data were drawn from the Brains for Dementia Research study.


Background
Multimorbidity, the presence of two or more health conditions, has been identified as a possible risk factor for clinical dementia.It is unclear whether this is due to worsening brain health and underlying neuropathology, or other factors.In some cases, conditions may reflect the same disease process as dementia (e.g.Parkinson's disease, vascular disease), in others, conditions may reflect a prodromal stage of dementia (e.g.depression, anxiety and psychosis).

Aims
To assess whether multimorbidity in later life was associated with more severe dementia-related neuropathology at autopsy.

Method
We examined ante-mortem and autopsy data from 767 brain tissue donors from the UK, identifying physical multimorbidity in later life and specific brain-related conditions.We assessed associations between these purported risk factors and dementia-related neuropathological changes at autopsy (Alzheimer's-disease related neuropathology, Lewy body pathology, cerebrovascular disease and limbic-predominant age-related TDP-43 encephalopathy) with logistic models.

Results
Physical multimorbidity was not associated with greater dementia-related neuropathological changes.In the presence of physical multimorbidity, clinical dementia was less likely to be associated with Alzheimer's disease pathology.Conversely, conditions which may be clinical or prodromal manifestations of dementia-related neuropathology (Parkinson's disease, cerebrovascular disease, depression and other psychiatric conditions) were associated with dementia and neuropathological changes.
2][3] However, the mechanisms of this are unclear.Multimorbidity may contribute to dementia risk by exacerbating underlying brain pathologies such as Alzheimer's disease, Lewy body disease or cerebrovascular disease.An alternative explanation is that factors associated with multimorbidity may predispose people to cognitive impairments from other causes, such as functional cognitive disorders. 4Longitudinal cliniconeuropathological studies provide an opportunity to directly test these associations between multimorbidity and pathology seen at autopsy.We tested whether autopsy data from the UK Brains for Dementia Research (BDR) programme supported a hypothesised link between multimorbidity and dementia-related pathology.

Method
Participants BDR participants were recruited from six sites across England and Wales (Newcastle, Manchester, Bristol, Cardiff, Oxford and London), and provided written informed consent for repeated research assessment and for brain tissue donation.Research visits were facilitated by an informant (e.g. a family member or close friend), where available, and were conducted every 1-2 years after baseline until death.Prospective participants were identified through local research studies and clinical services, public research participation events, newsletters and online advertisement.This cohort was restricted for analysis to those who died aged at least 60 years and had had at least one ante-mortem assessment to provide details of LTCs.Presence of dementia was ascertained through administration of the Clinical Dementia Rating (CDR) scale at each visit, and defined as a CDR global score ≥1.

Brain tissue donation
Brain tissue was donated post-mortem.Samples underwent standardised neuropathological assessment as previously described 5 to assess: (a) Alzheimer's disease-related neuropathological change, 6 rated by Thal phase of amyloid deposition, 7 Braak staging for neurofibrillary tangle (NFT) pathology, 8  Additional less-common pathologies were also assessed on a case-by-case basis, including argyrophilic grain disease, 13 corticobasal degeneration, 14 frontotemporal lobar degeneration 15 and Creutzfeldt-Jakob disease.Since these had low prevalence in this cohort they were not included as modelled outcomes.

Health data extraction
Data on LTCs were extracted from three complementary sources.ICD-10 codes were reported for each clinical diagnosis by BDR clinical research staff, incorporating all information available (clinical research and primary care records, where available).Responses to specific health questions were identified from the Cambridge Mental Disorders of the Elderly Examination (CAMDEX) interview, again rated by BDR-trained clinical staff.Finally, free-text responses to the CAMDEX medical history questionnaire were systematically searched to identify LTCs not reported elsewhere.In the case of disagreement between clinically rated conditions and CAMDEX-reported conditions, the former (ICD-10 code) was treated as the most informative source.Those without CAMDEX data, or with any missing answers to the CAMDEX health questionnaire, were excluded.A single report of any given condition was sufficient to consider this as being present, so long as this corresponded to a formal long-term diagnosis (e.g.major depressive disorder would qualify as a long-term condition, but depressive symptoms reported in psychological testing alone would not).

Defining multimorbidity
To enable stratification of groups by multimorbidity, key agerelated LTCs from the Charlson Comorbidity Index (CCI) were identified, with ICD-10 codes corresponding to previous research. 16odifications and supplements were made to the standard CCI to enable appropriate group comparisons.To prevent circular reasoning, diagnoses of clinical dementia (included in the standard CCI) were entirely excluded from multimorbidity classification.
2][3] We therefore also sought information on the presence of these conditions in addition to the CCI measures to test how the inclusion of these conditions affects the association between LTCs and dementia-related neuropathological change.
LTCs which could be clinical or prodromal manifestations of dementia-related neuropathological changes (Parkinson's disease; cerebral haemorrhage, infarct, stenosis or other cerebrovascular disease; depression or other psychiatric condition) were not treated as indicators of multimorbidity in our primary analysis.These were instead grouped under a 'brain comorbidity' category and examined as separate predictors in secondary analyses.
Causes of death (e.g.fatal myocardial infarction) were not considered as indicators of multimorbidity, unless these had also been reported previously in life.

Analysis
Associations between multimorbidity and neuropathological changes were assessed with Bayesian logistic models, adjusting for random differences between sampling sites, age at death, and both with and without apolipoprotein E4 (APOE4) genotype for Alzheimer's disease-related changes (available only for a subset of cases).
Staged neuropathological changes (Thal phase, Braak tangle stage, CERAD score and overall VCING severity) were examined with ordinal models.Binary changes (Lewy body Braak stage ≥IV, LATE-NC, subcortical infarcts >10 mm, CAA and white matter arteriolosclerosis) were estimated with Bernoulli models, as was clinical dementia as an outcome, adjusting for age and education.
Models were estimated with the brms package for R software, as an interface to the Stan probabilistic programming language in Windows.Sensitivity analyses were undertaken with a range of flat, weakly informative and informative t-distributed priors, and with probit-link models to assess the robustness of findings.
Sampling of posterior parameter estimates was undertaken with the No-U-Turn Sampler.Four chains were run in parallel for 2000 iterations (1000 warm-up iterations) initially, with any nonconvergence or inefficiency of chains diagnosed and addressed as required by increasing the target acceptance probability or number of iterations.Models were then re-estimated with 6000 iterations to verify that convergence had been achieved.The effects of including APOE status were assessed in sensitivity analyses with missing data multiply imputed by Bayesian methods, which also assessed any effects of missingness in other variables.

Ethics
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.All procedures involving human participants/patients were approved by the Health Research Authority North East -Newcastle & North Tyneside 1 Research Ethics Committee (18/NE/0124).

Primary analysis: physical multimorbidity
In total, 767 participants had undergone autopsy and provided sufficient information to assess comorbid health conditions from the CCI; 328 were cognitively healthy or had mild cognitive impairment, 439 had clinical dementia based on ante-mortem assessment (i.e.without reference to pathological assessment).Overall, there was a mean interval of 4.0 years (s.d.= 2.8) between the first observation and death, although this was shorter in those with dementia.The mean age at recruitment into the BDR cohort is 75.9 years (s.d.= 8.5), although the available cohort with autopsy were older on average at initial assessment (Table 1).APOE status was known for 453 brain tissue donors, with 223 (49%) having one or more ε4 alleles.The most common physical conditions reported were cancer (n = 261), myocardial infarction and diabetes (both n = 98), with all other assessed conditions being relatively more uncommon (Table 2).
The majority of participants reported at least one LTC in addition to dementia (where present).There was a higher rate of cancer in those who were dementia free, and none of the primary long-term conditions were clearly more common in those with dementia than those without (Table 2).When examining multimorbidity as a possible moderator of the relationship between overall Alzheimer's disease-related pathology and presence of dementia, presence of multimorbidity weakened the relationship between the diagnosis of clinical dementia and the presence of Alzheimer's pathology.That is, in those with multimorbidity, clinical dementia was less likely to be associated with Alzheimer's pathology, compared with those without multimorbidity (Supplementary Fig. 1, available at https://dx.doi.org/10.1192/bjp.2024.25).

Secondary analysis: brain comorbidities
We conducted several secondary exploratory analyses to test the effects of including different indicators of multimorbidity which have been included in previous research.These brain comorbidity measures included conditions that may be clinical or prodromal manifestations of dementia-related neuropathological changes: Parkinson's disease and clinical diagnosis of cerebrovascular disease (which can directly cause clinical dementia) and psychiatric disorders, which can have direct cognitive effects or can be prodromal to dementia (depression and non-depressive mental health conditions such as anxiety and psychosis).Personality and stress disorders were also examined as in previous studies; there were no cases of personality disorder reported and a single case of post-traumatic stress disorder reported.
In contrast to the physical multimorbidity measures, multimorbidity of brain LTCs was clearly associated with substantially increased risks of dementia (Supplementary Fig. 2).This effect seemed to be driven primarily by Parkinson's disease, depression and other mental disorders: presence of any of these was associated with greater odds of dementia (OR = 1.8, 95% CI 1.2-3.0),which increased further as more predictors were observed (Supplementary Fig. 1(a)).
Examining the association with neuropathological changes, the individual and combined presence of Parkinson's disease, depression and non-depressive mental disorders were associated with increasing risks of Lewy body pathology, specifically as more of these conditions were observed (Supplementary Fig. 1(b)), and adjusting for presence of Lewy body pathology largely attenuated the association between these conditions and clinical dementia (OR = 1.49, 95% CI 0.85-2.58).These conditions were not evidently associated with Alzheimer's disease-related neuropathological changes or any measures of cerebrovascular pathology.

Psychiatric multimorbidity
Finally, we assessed whether excluding Parkinson's disease as an indicator of multimorbidity, while retaining depressive and nondepressive mental disorders, was sufficient to remove the association between dementia and Lewy body pathology.The association between mental health conditions and clinical dementia remained (OR = 1.74, 95% CI 1.03-2.99for presence of one; OR = 3.03, 95% CI 1.06-8.94for multimorbid depressive and non-depressive mental disorders; Supplementary Fig. 1(c)).The association with Lewy body pathology, however, was not strongly supported in the absence of Parkinson's disease (OR = 1.58, 95% CI 0.86-2.76for presence of one; OR = 2.48, 95% CI 0.73-7.61for psychiatric multimorbidity; Supplementary Fig. 1(d)), and there remained no clear relationship between these and Alzheimer's disease pathology or cerebrovascular pathology.
Although there was a reasonably strong separate association between dementia-related LTCs and dementia, directly incorporating these as indicators of multimorbidity was not sufficient to cause a positive association between overall multimorbidity and clinical dementia in this cohort as brain comorbidities were less common than physical comorbidities.
All analyses showed good convergence of sampling chains, with all R values <1.01 and sufficient effective sample sizes.Sensitivity analyses tested the influence of prior choices on the outcome, testing flat, weakly informative and informative priors (anticipating a positive association between multimorbidity and neuropathological change, consistent with previously reported associations with dementia).These analyses did not meaningfully change the results for any of the considered clinical or neuropathological outcomes, suggesting that the findings were not simply dictated by the influence of the prior, nor do they reflect a lack of data (in which case the informative prior would have the greatest influence); the data were robustly incompatible, with a positive association between primary multimorbidity measures and dementia-related neuropathological change.We also assessed the impact of missing pathological data (Supplementary Table 1) or APOE status with imputed data-sets, which similarly did not change any findings.
Additional sensitivity analyses sought to examine the robustness of the choice of link function: probit models provided similar results to those presented here, with slightly attenuated risk ratios but narrower confidence intervals.

Discussion
We tested whether multimorbidity would be associated with greater dementia-related neuropathology in this moderately sized UK cohort.We found no evidence of a positive association between physical multimorbidity and dementia-related neuropathological changes.Physical multimorbidity weakened, rather than strengthened, the association between clinical dementia diagnosis and Alzheimer's disease-related pathology.In contrast, the occurrence or co-occurrence of specific LTCs that may be clinical or prodromal manifestations of dementia-related pathology -Parkinson's disease, cerebrovascular disease, depression and other psychiatric disorderswas positively associated with rates of clinical dementia and corresponding Lewy body pathology.
This does not support the hypothesised link between overall multimorbidity and dementia-related pathology such as Alzheimer's disease, and suggests that mechanisms other than increasing dementia-related pathology may account for the reported relationship between overall multimorbidity and clinical dementia (Fig. 2).
Key considerations for interpreting our findings in the research context include the selection of appropriate indicators of multimorbidity for dementia risk prediction, the differentiation of sustained, progressive dementias from transient cognitive complaints, the presence of cognitive symptoms as a direct consequence of illness, and the possible role of primary psychiatric conditions.

Selection of multimorbidity indicators in dementia
Multimorbidity is not operationalised in a consistent manner across studies.Brain comorbidities (Parkinson's disease, stroke/cerebrovascular disease and primary psychiatric disorders) have previously been treated as risk factors for dementia alongside physical LTCs. 1 This may be problematic, as brain comorbidities such as these have a different causal relationship with both dementia and its associated pathologies, being brain conditions and in some cases (Parkinson's disease, cerebrovascular disease) caused by dementia-related neuropathologies.
Consistent with this, we directly assessed brain multimorbidity separately and found that, unlike physical (non-brain) multimorbidity, this had a positive relationship with clinical dementia and associated pathology particularly owing to the inclusion of Parkinson's disease.In dementia risk factor studies, inclusion of Parkinson's disease alongside other multimorbidity measures is likely to confound findings, given that the Lewy body disease underlying this is also a dementia pathology.

Diagnosis of dementia across settings
The diagnosis of clinical dementia does not necessarily reflect the presence of neurodegenerative or cerebrovascular disease: acute cognitive impairments, psychiatric disorders and functional cognitive disorders may mimic neurodegenerative dementia. 4isdiagnoses are known to occur, with dementia diagnoses sometimes being rescinded.The number of dementia diagnoses in healthcare settings therefore reflects the sum of two inputs: the number of progressive dementias and the number of potentially reversible dementias (Fig. 2).Which of these numbers is being modulated by any theorised risk factor (such as multimorbidity) is not always clear, and may require deliberate research designs to examine.
The cohort described here benefited from longitudinal follow-up with objective reassessment of cognitive function.We are therefore reasonably confident that dementia diagnoses correspond to sustained, objective impairments.Relatedly, large and population-representative studies with repeated assessment of objective cognitive dysfunction have not supported an association between several physical LTCs and progressive cognitive impairments. 17t has been common for large risk factor studies to not objectively assess (and subsequently reassess) dementia, instead deriving this outcome from healthcare records: for example, by examining the first reported onset of dementia 1 or seeking records of dementia-related healthcare claims. 2This may raise the risk of including dementia cases with only a transient cognitive impairment alongside those with a progressive dementia.Such transient or non-progressive dementias will likely have a different aetiology, reflecting causes other than progressive underlying brain pathology.As discussed below, physical and mental factors associated with multimorbidity could be direct causes of transient or non-progressive cognitive symptoms.This might better account for previously reported associations between multimorbidity and dementia in the absence of greater neuropathological change.
Improving the recognition and understanding of such potentially reversible dementias, and any possible links to psychiatric and physical comorbidities, is crucial for future dementia research.Such cases may be present in observational and interventional research studies, particularly those without biomarker or neuropathological confirmation of disease, with important implications for statistical power and interpretation of results.

Cognitive symptoms and physical comorbidities
We found that physical multimorbidity had a moderating effect of weakening the relationship between Alzheimer's disease pathology and clinical dementia; these results are similar in direction and magnitude to reported moderating effects of frailty, a related concept. 18his effect appeared to be driven by an under-representation of Braak NFT stages V and VI (when there is a high likelihood of cognitive symptoms due to Alzheimer's disease) in people with physical multimorbidity and an over-representation of Braak NFT stages I and II (when Alzheimer's disease-related changes have a very low likelihood of causing cognitive symptoms).
This may be explained by the acute or chronic illness directly impairing cognitive performance, mimicking neurodegenerative dementia in the absence of significant pathology.There are several direct consequences of physical multimorbidity which may predispose people to experiencing cognitive symptoms in the absence of dementia-related pathology.In cognitively healthy older adults, physical multimorbidity is associated with greater prevalence of subjective cognitive symptoms 19 an association mediated by stress, poor sleep and anxiety.Pain and fatigue, possible consequences of multimorbidity, may also be associated with a profile of cognitive symptoms characteristic of functional cognitive disorders. 20Polypharmacy is a natural consequence of multimorbidity, with multiple LTCs requiring multiple overlapping treatments.There is a well-recognised association between polypharmacy and cognitive symptoms in later life, particularly when there is an increasing anticholinergic burden. 21ubjective, functional or transient objective cognitive symptoms related to physical comorbidities and polypharmacy could therefore contribute to an increased number of cases with dementia diagnosis in healthcare records. 1   Associations between multimorbidity and neuropathology in dementia other settings, and would not be associated with underlying neurodegenerative pathology, potentially explaining divergent findings to date.
We did not find an association between physical multimorbidity and key markers of neuropathological change in this cohort.However, several pathobiological mechanisms could contribute to pathology-related change without being reflected in these neuropathological findings.Synaptic dysfunction/loss, neuroinflammation, mitochondrial dysfunction and cerebral hypoperfusion/ hypometabolism are possible contributors to cognitive dysfunction which may not be reflected by neuropathological staging.Any of these could be important unmeasured mediators between physical multimorbidity and cognitive impairment, requiring further examination.

Cognitive symptoms and psychiatric comorbidities
Psychiatric comorbidities such as depression may be prodromal features of dementia-related neuropathology, but may also mimic dementia-like cognitive symptoms. 22This could partially account for previously reported findings of a link between overall multimorbidity (with previous studies often including mental illnesses) and dementia.
Unlike physical multimorbidity, we found psychiatric multimorbidity to be positively associated with clinical dementia.When co-occurring with Parkinson's disease, this was explained by underlying Lewy body disease.In the absence of Parkinson's disease, however, this pathological link was not clear.This would be consistent with the dual nature of psychiatric comorbidities as both manifestations and mimics of dementia-related brain changes (Fig. 2).Given the absence of a clear link here between isolated mental health conditions and dementia-related pathology, the observed link between these and dementia seemed mostly unrelated to these being prodromal manifestations of neurodegeneration.This could also suggest a lack of support for hypothesised psychiatriconset Lewy body disease.However, owing to likely heterogeneity this warrants further, detailed assessment.We assessed any reported history of long-term psychiatric conditions: neuropsychiatric and behavioural symptoms of dementia may not necessarily result in such a long-term diagnosis.
Although individuals with a cognitive disorder secondary to a psychiatric disorder should not meet consensus criteria for allcause dementia, 23 misdiagnosis is common.Misdiagnosis of cognitive symptoms in primary psychiatric disorders, or the prodromal manifestation of psychiatric disorders in developing degenerative disease, could therefore partially explain the apparent link between multimorbidity and dementia in this and previous studies.This may be particularly pertinent in younger dementia cohorts (i.e.individuals in their 60s and 70s), when differentiation of dementia from mood disorder is less accurate.

Strengths and limitations
We used data from a clinicopathological study benefiting from comprehensive neuropathological assessment, providing gold standard evidence of the presence/absence of dementia-related neuropathology, and drawing from multiple sampling sites to cover regions across England and Wales.This included prospective follow-up of people with dementia and controls.Although the overall number of participants is smaller than most large epidemiological studies, the number with dementia and relative confidence in their diagnoses is a strength.
Drawing an older sample from clinical services and research cohorts, individuals with dementia in the BDR cohort may have a higher expected prevalence of neurodegenerative changes in contrast to younger population studies.Presence of dementia was assessed through administration of the Clinical Dementia Rating (CDR) scale within the study by experienced clinical researchers; this is a limitation of this work as the global CDR score is not a diagnostic scale.Final clinicopathological diagnoses were made by an expert clinical panel, but these were not used in this analysis, to limit bias from inclusion of post-mortem findings in ante-mortem clinical ratings.Although clinical dementia diagnoses may not accurately identify dementia subtype in this cohort, they have previously been shown to be generally accurate as to the presence of dementia-related neuropathological changes overall. 24his study was primarily designed to test the association between physical multimorbidity and dementia-related neuropathological change.Detailed assessment of the relationships between multimorbidity, polypharmacy, functional cognitive symptoms and dementia would require carefully designed studies for this specific purpose.Our above explanations are therefore consistent with the data available, but require testing in future studies.
This study did not have data linkage to electronic health records, and multimorbidity was calculated primarily through self-report, supplemented by clinical assessment, with a focus on key agerelated diseases.Comorbid conditions could therefore be missed, if not included within the CCI measure, or through not being reported by those with a more severe cognitive impairment (although informants or carers were also interviewed where available).
Individuals who volunteer for prospective research and future brain tissue donation are likely to be healthier than the wider population, which may be a source of bias.Although cancer, myocardial infarction and diabetes were common, other conditions were not, potentially limiting statistical power.Comparable population-representative cohorts have reported higher rates of multimorbidity than found here, 25 although they included different indicators of multimorbidity (e.g.hearing impairment).Although not fully population-representative, BDR participants came from multiple geographical regions with varying levels of deprivation, 5 which may somewhat attenuate the typical research bias.
Clinically reported cerebrovascular disease was relatively common in both cognitively impaired and unimpaired groups, contrary to expectations.This may reflect the heterogeneity inherent in cerebrovascular disease as assessed here (which includes strokes, transient ischaemic attacks and other cerebrovascular events), as well as the poor concordance between clinical and pathological assessment of cerebrovascular disease.
In contrast to previous research, we found no evidence that physical multimorbidity was associated with clinical dementia.However, the majority (56%) of donors with dementia also had one or more comorbid physical LTCs and therefore met broader criteria for overall multimorbidity, since dementia is itself a serious LTC.These comorbidities are likely to affect quality of life and care in dementia, even if they do not contribute to worse dementia-related neuropathology.
With an average of 4 years of follow-up before death and an average age at death between 80 and 89, these data represent associations of later-life multimorbidity, albeit the presence of these morbidities can reach back to earlier life.Previous studies assessed mid-life multimorbidity directly and found this to have a stronger relationship with dementia than late-life multimorbidity. 1 It is therefore possible that these associations shift over time, as both multimorbidity and neurodegeneration become more common with increasing age.This is further complicated by possible survivorship bias: those who develop dementia in later life have not died of another cause earlier, which might induce an apparent negative association where no association exists.Future research including neuropathological assessment may therefore benefit from more comprehensive assessment of comorbid conditions, particularly including their historical presentation, through health record linkage.

Psychiatry in music
An empty bliss beyond this worldthe music of The Caretaker as a representation of dementia The Caretaker is a musical project by English artist James Leyland Kirby running from 1999 to 2019 that has garnered critical acclaim for its exploration of themes related to memory and nostalgia.The music of The Caretaker is particularly notable for its links to dementia and its representation in unique fashion.
Through his use of samples of old, degraded music and sounds, often from ancient records he found, The Caretaker's music represents the process of degradation mirrored in dementia.The links are evident from the project's earliest releases, indeed one of the first bears the title 'Theoretically pure anterograde amnesia'.These samples are then distorted, fragmented, looped and processed creating a sense of disorientation and confusion that mirrors the experience of living with dementia.
In addition, the music is often characterised by a sense of nostalgia, imbued from the maudlin source material he has sampled.This nostalgia reflects the way in which people with dementia may cling to long-term memories from their past as their present experiences become increasingly confusing and disorienting with the music representing these periods of lucidity.
One of The Caretaker's most evocative albums is 'An empty bliss beyond this world'.It features samples of 1930s ballroom music, which have been distorted and manipulated, to create a sense of dislocation and decay.The music is often interrupted by bursts of static, suggesting the breakdown of communication, disorientation and confusion that is characteristic of dementia.The album title reflects the stage where the patient is beyond human contact.The album art comes from a painting entitled 'in spite of' by Ivan Seal, from a series named by psychiatrists and psychologists without pre-information.Ivan Seal's works are closely interlinked with The Caretaker's.
The most ambitious, and final project of The Caretaker was a six-album cycle called 'Everywhere at the end of time' released between 2016 and 2019.Conceptually this took the project to a logical conclusion by giving The Caretaker character himself dementia and the musical representation of this.This was achieved via similar means of utilising, at first quite pleasant samples of old records but over the cycle they become increasingly distorted, manipulated and indeed destroyed, with the final album representing end-stage dementia and death.
It is an uncomfortable listen and almost overwhelming, but it is a magnificent achievement and almost sublime.The apotheosis of an extremely important corpus of work of this century that I think about often.

Fig. 2
Fig. 2 Theorised pathways by which subcategories of multimorbidity might result in greater rates of dementia diagnosis.Solid lines indicate pathways supported by presented data, dashed lines indicate theorised explanations which could remain consistent with previous research findings.
Adam FlynnI spent 6 months as a psychiatry trainee in The Waterside Hospital, Derry/Londonderry in Northern Ireland.My rotation was in psychiatry of old age, primarily on the dementia ward.Of note, their occupational therapy room was covered in very large photos from Ireland's showband era of the 1950s, harking back in a nostalgic way to a culture that has long been lost.This made a strong impression on me in terms of thoughts about what constitutes memory, how it changes and degrades over time, and linked my mind to a music project by an artist called 'The Caretaker'.
Image: 'in spite of' 40 × 30 cm oil on canvas.Copyright Ivan Seal 2010.Reproduced with permission from the artist.© The Author(s), 2024.Published by Cambridge University Press on behalf of Royal College of Psychiatrists The British Journal of Psychiatry (2024) 224, 244.doi: 10.1192/bjp.2023.53

Table 1
Demographics of sample, stratified by cognitive status IQR, interquartile range; LTC, long-term health condition.

Table 2
Rates of each reported long-term health condition, stratified by cognitive status These are likely to not manifest as progressive cognitive impairment sufficient for dementia diagnosis in 22 Rock PL, Roiser JP, Riedel WJ, Blackwell AD.Cognitive impairment in depression: a systematic review and meta-analysis.Psychol Med 2014; 44: 2029-40.23 McKhann GM, Knopman DS, Chertkow H, Hyman BT, Jack CR, Kawas CH, et al.The diagnosis of dementia due to Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association Workgroups on diagnostic guidelines for Alzheimer's disease.Alzheimers Dement 2011; 7: 263-9.24 McAleese KE, Colloby SJ, Thomas AJ, Al-Sarraj S, Ansorge O, Neal J, et al.Concomitant neurodegenerative pathologies contribute to the transition from mild cognitive impairment to dementia.Alzheimers Dement 2021; 17: 1121-33.25 Bennett HQ, Kingston A, Lourida I, Robinson L, Corner L, Brayne CEG, et al.The contribution of multiple long-term conditions to widening inequalities in disability-free life expectancy over two decades: longitudinal analysis of two cohorts using the Cognitive Function and Ageing Studies.eClinicalMedicine 2021; 39: 101041.