Discussion

The development of memory services in England as a bespoke pathway for assessment of people with suspected dementia, separate from community mental health and neurology services, was promoted by the 2009 National Dementia Strategy.³ The subsequent expansion of memory services was supported by an optional quality improvement infrastructure (Memory Services National Accreditation Programme) led by the Royal College of Psychiatrists. This was not accompanied by a national audit data-set, in contrast to the National Audit of Dementia, which examines care in acute hospitals. Notably, national surveys of memory services conducted in 2013 and 2014 did not include patient-level data. This first large-scale audit of memory services in England provides important insights into diagnostic and immediate post-diagnostic practices, including how these differ by age.

Our data show that only a small proportion of younger people referred to memory services have dementia (14% of those aged <55, 23% of those aged <65), in accordance with our previous audit round in London and findings from Cambridgeshire.^{Reference Cook, Nichol and Isaacs2,Reference Yeung, MacFarland, Mlilo, Dean and Underwood4} English memory services diagnose 1 person with YOD for every 40 with LOD. This makes YOD a small fraction of memory service workload and might limit the ability of memory service clinicians to build expertise in this condition, which the Royal College of Psychiatrists has identified as a priority.⁵ Whether younger people are better served by specialist YOD services or can be appropriately supported within ageless memory services remains a subject of debate.^{Reference Yeung, MacFarland, Mlilo, Dean and Underwood4,Reference Stamou, La Fontaine, Gage, Jones, Williams and O'Malley6} Either way, the higher prevalence of psychiatric disorders in younger people referred to memory services suggests that psychiatry-led services are an appropriate initial point of referral provided they include cognitive neurologists or have good links to specialist neurological services.

Do the audit data suggest that memory services are underskilled in YOD and rarer dementias? Population studies suggest that FTD accounts for 10% of YOD and just under 3% of LOD.^{Reference Hogan, Jetté, Fiest, Roberts, Pearson and Smith7} Our finding that 7% of patients with YOD and 0.3% of patients with LOD were diagnosed with FTD likely reflects either referral bias or underdiagnosis.

Younger individuals presenting to primary care with atypical symptoms such as language decline or personality change might be referred to neurology services or community mental health services and therefore not appear in memory service cohorts. Nevertheless, pathological series suggest that over half of FTD cases might be clinically misdiagnosed, most commonly as Alzheimer's disease.^{Reference Selvackadunco, Langford, Shah, Hurley, Bodi and King8}

In the absence of disease-modifying therapy, the impact of subtype misdiagnosis on disease progression is relatively limited; nevertheless, dementia subtypes differ in terms of aetiology, prognosis and response to symptomatic treatment. Counselling advice given to patients and families is therefore potentially misleading if based on an inaccurate subtype.

Furthermore, nearly one in ten individuals with YOD in this sample were diagnosed with ‘unspecified dementia’, compared with 6% of those with LOD. Dementia in younger people is more likely to be caused by a single neuropathological process and confounding factors such as frailty are less prevalent. A specific subtype diagnosis should be attempted wherever possible, particularly given the higher rate of genetic mutations in YOD, so that appropriate counselling can be given to patients and families.

The audit demonstrates the resource implications of assessing individuals with YOD, with higher proportions of these patients being referred for diagnostic neuropsychology and MRI scans. In the audit organisational questionnaire, 87% of services reported that they could refer patients for DAT scans, 77% for PET scans and 56% for CSF examination. However, the low rate of functional and biomarker-based investigations in both younger and older patients suggests that these referral pathways are rarely used. Routine use of biomarkers in dementia diagnosis remains contested.^{Reference Isaacs and Boenink9} Nevertheless, our data suggest a system largely unfamiliar with molecular diagnosis in dementia and in need of significant additional educational and financial resources should this become necessary.^{Reference Ritchie, Russ, Banerjee, Barber, Boaden and Fox10}

That 31% of patients aged <65 and one in four of all patients who underwent neuropsychological assessment did not receive any diagnosis suggests that opportunities for therapeutic intervention may be being missed. Although not everyone referred to secondary care has a cognitive or psychiatric disorder, that people choose to attend memory service assessment suggests that they have at least some unmet need. A recent systematic review reported that people seeking help for subjective cognitive symptoms have high rates of psychological morbidity and should not be dismissed as ‘worried well’.^{Reference McWhirter, Ritchie, Stone and Carson11} It is possible that patients coded in the audit as ‘no diagnosis given’ received a diagnosis after the audit census date. However, given that services completed the audit 4–8 months after patients were first seen, this would suggest that a significant minority of memory service patients, particularly those aged <65, are waiting many months for a diagnosis.

Although it is important not to pathologise people who present with mild symptoms not meeting thresholds for psychiatric diagnosis, giving meaning to people's symptoms usually requires healthcare professionals to give those symptoms a label or to provide an explanatory formulation. That this is often not possible even following a detailed neuropsychological assessment suggests that the framing of assessment in memory services as positioning patients on a deterministic pathway from ‘normal’ to MCI to dementia risks missing non-dementia contributors to cognitive symptoms.

Functional cognitive disorder (FCD) is an increasingly recognised cause of distressing cognitive symptoms.^{Reference Ball, McWhirter, Ballard, Bhome, Blackburn and Edwards12} The use of this diagnosis for only 5% of patients aged <65 and 0.1% of those aged ≥65 in our sample suggests underrecognition or coding issues. FCD does not feature in the UK edition of SNOMED CT, the coding catalogue used by NHS Digital; the closest match is ‘Dissociative neurological symptom disorder co-occurrent with cognitive symptoms’, which memory service clinicians are unlikely to use, as ‘dissociation’ is rarely the lead symptom in FCD.

The key features of FCD can only be elicited though careful clinical observation;^{Reference Poole, Cope, Bailey and Isaacs13} it cannot be ruled in by investigations such as brain imaging or cognitive instruments. Interestingly, use of FCD as a diagnostic label was associated with a lower scanning rate, suggesting that its increased use might help reduce over-investigation.

Similarly, the proportion of younger people in this sample given a diagnosis of MCI seems too high, given that MCI was developed to define people considered at high risk of progressing to dementia, rather than as a generic term for cognitive symptoms not meeting threshold for dementia diagnosis.^{Reference Dunne, Aarsland, O'Brien, Ballard, Banerjee and Fox14} Where used inappropriately, MCI risks patients being given an inappropriately adverse prognosis and missing out on personalised psychotherapeutic interventions for FCD and psychiatric causes of cognitive symptoms. Conversely, lack of confidence in diagnosing dementia in younger people might lead to inappropriate use of MCI as a temporising measure, which has the effect of delaying diagnosis and intervention. This would be particularly injurious to younger people with dementia, who generally endure longer times from symptom onset to diagnosis than older people.^{Reference Van Vliet, De Vugt, Bakker, Pijnenburg, Vernooij-Dassen and Koopmans15} In our view, MCI is most appropriately used not in isolation but as part of a formulation, in which likely contributory factors are cited.

In this audit, waiting times were similar for both younger and older people diagnosed with dementia. However, the audit only looked at waiting times following referral from primary care rather than from symptom onset. Longer duration from onset to diagnosis in YOD could still be caused by delayed symptom recognition and referral into secondary care.

Although our sample of people with YOD was relatively small, some potentially useful epidemiological observations emerge. People with YOD were nearly three times more likely to be smokers than those with LOD. Smoking prevalence among those with YOD was higher than in the general population.¹⁶ Smoking is a known risk factor for dementia,^{Reference Livingston, Huntley, Sommerlad, Ames, Ballard and Banerjee17} and vascular disease in general increases the risk of YOD.^{Reference Cations, Withall, Low and Draper18}

Alcohol consumption among individuals with LOD in our sample was broadly in line with population data, but was higher among those with YOD than in the general population,¹⁹ reflecting the known association between heavy alcohol use and dementia.¹⁵

In a quarter of cases, CST was not available for people with YOD, and in services where it was offered it was felt not to be appropriate for 54% of patients. However, when CST was offered to those with YOD, only 17% declined, compared with 45% in the LOD group. More work is required to understand why CST is either unavailable or felt to be inappropriate in such a large proportion of the YOD population. It is possible that memory service clinicians assumed that the content of the CST sessions would not be age appropriate for the younger population. Some services offer cognitive rehabilitation or occupational therapy to support functional ability in working-age patients. YOD-specific peer support groups are also available in some areas. These activities may be preferred to standard CST offered by memory services, which is often aimed at older people. However, clinicians should not rule out the possibility that a younger patient might accept CST.

Carer psychoeducation was not available for just over a third of individuals with YOD. Where offered, only 13% declined it, compared with 30% in the LOD group, although the small number in the YOD group meant that this difference did not reach statistical significance. There may be an assumption that the carer of a person with YOD might have time constraints related to being in employment or other caring responsibilities that preclude engagement. However, there may be an increased need for support among younger carers precisely because they typically have existing commitments and because of the extreme emotional, practical and financial dislocation associated with YOD.³

There were no discernible differences between YOD and LOD with respect to acceptance of care navigation. However, clinicians report difficulties finding age-appropriate resources and people with YOD and their carers report that services are unable to meet their personal and psychological needs.^{Reference Cations, Withall, Horsfall, Denham, White and Trollor20}

That people with LOD were less likely to accept evidence-based post-diagnostic interventions suggests that these are not always offered in a user-friendly way to the older population. Barriers to older people engaging in CST and carer psychoeducation include limited accessibility, travel costs, frailty or lack of a carer to accompany the patient. Commissioners and providers must ensure that interventions for which there is clear evidence of clinical and cost-effectiveness, and which are endorsed in national guidelines, are made available in ways that suit the needs of the population in whom they are indicated.

The audit results indicate that people with YOD are more inclined to engage with research participation than those with LOD. This likely reflects their greater mobility, as research centres are often regionally based. However, it also reflects the general underrepresentation of older patients in biomedical dementia research. This is problematic given that people aged 80 and over form the majority and the fastest growing cohort of those with dementia.