International consensus on quality indicators for comprehensive assessment of dementia in young adults using a modified e‐Delphi approach

To develop guidance for clinicians about essential elements that can support clinical decision‐making in the diagnostic workup of young onset dementia.


| Young onset dementia and diagnosis
Young onset dementia (YOD) refers to dementia diagnosed in those aged 65 years and under. The differential diagnosis of YOD encompasses complex presentations of the common primary neurodegenerative diseases as well as autoimmune, inflammatory, late onset metabolic and hereditary/familial causes. 1 While Alzheimer's disease (AD) makes up the majority of cases in younger adults, it represents a significantly smaller percentage than in late onset disease (LOAD) 2,3 and presentations are generally non-amnestic in nature. 4 Because of the complexity of presentation and often atypical nature of presenting symptoms, YOD is poorly recognised and often misdiagnosed 5,6 and advice to support clinicians with identification and assessment of key symptoms in the diagnostic workup is lacking. In particular, clinicians with less familiarity of dementia in younger adults can be unaware of red flag symptoms and essential investigations that can help identify the complex subtypes of dementia which are more common in younger people. Furthermore, routine assessments tailored to older patients are often insufficient to identify the significant overlap between psychiatric disorders and neurodegenerative disease in this age group and this can lead to delay in specialist referral, clinical under-investigation, misdiagnosis, and delays in obtaining a definitive diagnosis reference (7)(8)(9).
Given these concerns, a UK-based study, called The Angela Project, aimed to develop guidance on best practice in diagnosis. An indepth scoping review of the literature highlighted 29 papers identifying red flags in the clinical approach to diagnostic assessment of YOD and concluded that a clinically rigorous and systematic approach is necessary inorder to avoid misdiagnosis or under diagnosis for younger people with dementia. 7 To further a systematic approach to diagnosis in YOD, the present paper reports the findings from an international Delphi study with secondary care clinical experts, that identified key elements that support clinical decision-making. The Delphi method was adopted as it is an appropriate method for exploring clinical decision-making by consensus.
Our objectives were to explore the (a) key indicators essential for high quality assessment and diagnosis of YOD, and to (b) identify representative opinion from clinicians in a range of disciplines typically involved in diagnosis of YOD.

| Delphi method
The Delphi process 10,11 is an expert consensus method that can be used to develop best practice guidelines using practice-based evidence. This process comprises a series of structured surveys and is used to collate opinions on a set of matters in order to gain a consensus of opinions. 11 In diseases where clinical evidence is lacking it is a method deemed suitable for the development of guidelines on diagnosis and management and is often used in healthcare decisionmaking. 12 The goal is to translate professional experience into informed judgement and to support effective decision-making with an emphasis on stability of group opinions rather than individual opinions. The Delphi approach allows anonymised individuals to freely express their opinions, reconsider them in the light of collective opinions from the whole group and, with each round, gain consensus. 13 The Delphi process used here to determine consensus about key elements in the assessment and diagnosis of YOD involved four steps:

| Panel member selection and participants
Purposive sampling was used to select 52 experts who had a specific interest (such as through research or through their own practice) in YOD and these experts were approached to take part in a Delphi consensus study on diagnosis of YOD (UK -N = 28, Female = 4, Male = 24; International -N = 24, Female = 7, Male = 17). Criteria for experts were that they were (a) affiliated with a YOD service (b) were authors from papers found through the scoping literature review previously International (N = 8), Female = 5, Male = 3). Please see Table 1 for the demographics of the Delphi experts.

Key points
• Delay to diagnosis, misdiagnosis as a psychiatric condition and under-investigation of young onset dementia is common because of a lack of expertise.
• Routine assessments tailored to older patients are insufficient to identify the complex presentations usually seen in young people with dementia • A Minimum and Gold Standard set of indicators for high quality assessment of young onset dementia have been derived using a Delphi consensus study with international experts.
• The standards provide a tool to aid clinical decision-making for those with less experience in the field. Experts were approached by means of email and were sent a "Delphi invitation". Consent was considered implicit by completing the first round of the Delphi.

| Survey and rounds
The Delphi questionnaires were made available using the Bristol Online Survey (BOS) which allowed easy access for all the experts. A preliminary pilot study was conducted with a team of independent experts from a leading dementia research institution to provide initial feedback on the clinical vignette which featured in the first round on the survey.
The study was conducted in three phases spread over the period uk) on a Likert scale of 1-7, ranging from "not important at all", to "absolutely essential" (see Table 2).
Delphi Round 3: The statements that had overall mean scores below 6 (indicating they were moderately important or not important) and/or did not reach consensus in Round 2 were re-presented to the experts in Round 3. Experts were asked to re-read and reconsider their scoring if they wished. To provide additional decision-making support to the experts, given the variety of clinical specialisms, we provided the mean and standard deviation (SD) of the whole expert group per statement, as well the mean and SD per specialism (e.g.  Table 3 for an example of grouped free text quotes about mood). The items were further collapsed and grouped into universal descriptions to create a final list of 72 statements for rating.

| Round 2
Of the list of 72 unique statements presented in the second round (see Appendix B), 43 of the statements reached 80% consensus after Round 2, meaning they were rated in the upper threshold with scores of "absolutely essential" and "very important". Of these 43 statements, 15 statements were rated by all experts as being absolutely essential or very important, and a further 28 statements were rated by 80% of the experts as being "absolutely essential" or "very important". Please see Table 4 for the final list of statements that met consensus and Appendix C for statements that did not reach consensus after all rounds.

| Round 3
In total there were 29 statements where 80% consensus was not reached, or the statements were ranked as moderately important or less after Round 2. These were sent back to the experts for reconsideration in Round 3. In this round, experts were provided with the personal rating given for the statement in the previous round (Round 2),  Table 5).
As a result, the final list of statements following the three iterative rounds consisted of 48 statements where 80% consensus was reached that they were "absolutely essential" or "very important" to diagnosis in a younger adult. (see Table 4).
At the outset of the study, consensus was agreed as achieved on an item if at least 80% of the respondents were in agreement and the composite score fell in the upper threshold, defined as scores in the "absolutely essential" or "very important" range on the 7-point Likert scale.
Earlier studies have also used the certain level of 80% agreement to identify very high levels of consensus. 14 Obtaining a collateral history was considered to be an essential component of history taking mentioned by 100% of experts, in particular, noting any discrepancy between patient self-report and that of a story has elements to suggest an affective disorder, but this should not be taken at face value, especially without prior psychiatric history, and the degree of self-neglect. The ACE3 pattern is not suggestive of problems secondary to an affective disorder, for example, the low visuospatial scores, and language deficits (1013).  Experts reached consensus about the value of self-or observer-rated scales of mood or behaviour, but did not agree on which tool was most appropriate. It may be that whilst experts recognised the importance of a mood inventory, they may not have had the "discipline specific" expertise to call judgement on which specific tool is most suitable.
Consensus was reached on the value of obtaining further background information about biography and family history of dementia. Consensus about the value of access to quantitative volumetric analysis of medial temporal atrophy as a valuable biomarker in this age group was not reached, although assessment of medial temporal lobe atrophy by visual analysis was not.
In the event of normal baseline imaging, the experts were asked to rate which further investigations would be considered helpful. Cerebrospinal fluid (CSF) analysis for Tau and Amyloid biomarkers was rated by 15 experts as "absolutely essential" or "very important" with none rating it as of low importance although it did not reach consensus. Comparatively, Amyloid PET was rated as "absolutely essential" or "very important" by nine experts and four rated it as of "low importance" or "not at all important". There was general agreement that HMPAO SPECT was of limited value especially if FDG-PET was available.

| DISCUSSION
Using data from a three-stage modified Delphi study, consensus was Alzheimer's disease using conventional cognitive testing maybe unhelpful. In line with best practice in imaging, 15,16 a defined MRI dementia protocol including as T1, T2, and T2 FLAIR images was supported as was use of visual inspection of medial temporal lobe regional atrophy to discriminate Alzheimer's disease.
Consensus regarding the importance of mental state examination and the value of self-or observer-rated scales of mood or behaviour is designed to mitigate the high rates of psychiatric misdiagnosis of YOD, particularly as depression, identified in the literature. 19 Despite the potential advent of disease modifying treatments and the value in identifying prodromal dementia and high-risk populations, most likely to be those with YOD, no consensus was reached about the role of molecular biomarkers for diagnosis.

| Strengths and limitations of the study
In order to ensure that opinions were not biased, the researchers adopted a rigorous approach maintaining anonymity throughout to allow frank discussion. Written feedback about decision-making was encouraged to limit bias and avoid preconceptions. However the lack of representation of allied health professionals, such as occupational therapists and speech and language therapists, in the expert panel is likely to have influenced outcomes, although medically trained professionals are more likely to be directly involved in the final diagnostic decision. This is relevant to the Delphi results as working in a multidisciplinary environment with access to other key professionals was considered an important criterion by the participants.
As with all Delphi studies, the views expressed represent those of experts in specific disciplines and do not necessarily reflect the views of all experts in that field. In our study, the range of experts who agreed to participate may indicate bias in the selection process with those who have strong views being more likely to participate. As the sample was varied in terms of expertise in secondary care, care was taken to ensure that statements of importance by certain professional groups were not overlooked due to a lower number of experts from their discipline by providing experts with the "discipline" average score as well as the overall average when reviewing scores in Round 3. Thus, a statement not making the final list may have been due to lack of representation from a professional group in the expert panel, rather than it not being important to the diagnostic workup.
Whilst there is possibility of bias around particular concepts of assessment, we selected experts from diverse institutions who are highly qualified in the field, and conducted the survey in a rigorous manner by maintaining the anonymity of all participants to limit potential bias. The free text feedback was used to ensure that the structure and content of the survey did not impose preconceptions and experts were able to comment freely. Feeding back the scores of specialists from other disciplines on the statements facilitated re-con- Although steps have been taken to conduct the study in a rigorous manner, the template outlined is provided for guidance only and not as a definitive tool. It is hoped that it may help improve standards and provide a clinically useful tool, particularly for those with less experience in the field.
Guidelines in themselves may not ensure change in supportive behaviours. 10 Therefore, our next steps are to use these standards to explore current practice in the UK through a clinical case note audit of diagnosis of YOD using a digital platform. The goal is to identify compliance with the quality indicators in mental health settings across the UK to assess current clinical practice and to identify potential barriers and facilitators to high quality assessment. Mr Smith, a 60-year old, right handed retired accountant is referred to your clinic by his GP (Primary care physician US). He attends alone. His wife is concerned about his memory but she is at work and unable to attend. Mr Smith is a little dishevelled, and somewhat over familiar in manner. Mr Smith, does not feel he has any particular problems, says his wife is always nagging him and he sometimes worries she may be having an affair. He agrees that he sometimes forgets what he wants to say mid-sentence, has occasionally misplaced his keys and has lost interest in reading which he previously enjoyed. His golf buddies joke with him that he has poor head for numbers and they help him keep track of the scores. He has no significant past medical history, takes no regular medication and lives at home with his wife. His mother died in her 60s in a care home, with dementia. On further enquiry, Mr S reports that for the past 6 months, he has been sleeping poorly and admits that he worries about his son who is going through a difficult divorce. He still drives and manages the family accounts. Basic cognitive testing with the ACE-III demonstrates a total score of 77/100 with the following subset scores: Attention 16/ 18, memory 19/26, verbal fluency 10/14, VS 10/16, language 22/26.
With regards to this vignette, and making a diagnosis, please answer the following questions: Which are the key elements in the history that raise concern, if any? Could you outline them and explain why they are of note?
What further information would you require in order to refine your thinking?
What other tests might you undertake at this stage?
Please feel free to note any concerns or additional information you regard as important when diagnosing YOD: