Géraud Dautzenberg

Trust me, I’m a validated test!? Géraud Dautzenberg Unseen mild (cognitive) impairment and the use of the MoCA in an old age psychiatry setting.

Trust me, I’m a validated test!? Unseen mild (cognitive) impairment and the use of the MoCA in an old age psychiatry setting. Géraud Dautzenberg

Colofon The research described in this dissertation was conducted at and financially supported by the department of old age psychiatry of Altrecht, Utrecht the Netherlands. The dissertation was prepared at the department of psychiatry, Amsterdam UMC, VU university Amsterdam within the Amsterdam Public Health research institute. Printing of this dissertation has been accomplished with gratefully acknowledged financial support, provided by de Vrije Universiteit Amsterdam and Alzheimer Nederland (Amersfoort). ISBN: 978-94-6458-708-1 Cover: Roger Raveel - Nog een beetje mens. www.rogerraveelmuseum.be Cover-photo and -design: Jantine Messing / Coen van der Bijll Dissertation lay-out and design: Coen van de Bijll (Ridderprint/Publiss) Printing: Ridderprint, www.ridderprint.nl © Copyright 2022: Géraud Dautzenberg, The Netherlands

VRIJE UNIVERSITEIT Trust me, I’m a validated test!? Unseen mild (cognitive) impairment and the use of the MoCA in an old age psychiatry setting. ACADEMISCH PROEFSCHRIFT ter verkrijging van de graad Doctor aan de Vrije Universiteit Amsterdam, op gezag van de rector magnificus prof.dr. J.J.G. Geurts, in het openbaar te verdedigen ten overstaan van de promotiecommissie van de Faculteit der Geneeskunde op dinsdag 13 december 2022 om 11.45 uur in een bijeenkomst van de universiteit, De Boelelaan 1105 door Géraud Matthieu Félix Clément Dautzenberg geboren te Heerlen

promotor prof.dr. A.T.F. Beekman copromotor: dr. J.G Lijmer promotiecommissie prof.dr. R.W. Kupka prof.dr. A.W. Braam prof.dr. H. Weinstein prof.dr. W. Cahn dr. C.R. Tulner dr. A. Dols

TABLE OF CONTENTS Chapter 1. General Introduction Why, When, Where, Who, What, How ? 7 Section A: Unseen needs 43 Chapter 2. The Care Needs of Older Patients with Bipolar Disorder. 45 Section B: The MoCA validation in different old age settings 63 Chapter 3. Diagnostic accuracy of the Montreal Cognitive Assessment (MoCA) for cognitive screening in old age psychiatry: determining cutoff scores in clinical practice. Avoiding spectrum-bias caused by healthy controls. 65 Chapter 4. Clinical value of the Montreal Cognitive Assessment (MoCA) in patients suspected of cognitive impairment in old age psychiatry. Using the MoCA for triaging to a memory clinic. 85 Section C: The MoCA in clinical practice 109 Chapter 5. The Montreal Cognitive Assessment (MoCA) with a double threshold: improving the MoCA for triaging patients in need of a neuropsychological assessment. 111 Chapter 6. Severe cognitive impairment associated with a high free, but therapeutic total concentration of valproic acid due to hypoalbuminemia in an older patient with bipolar disorder. 137 Chapter 7. Summary of the main findings and general discusion. 151 Chapter 8. Appendix 195 Nederlandse samenvatting 196 Conversie Tabel voor ernst cognitieve stoornissen 200 List of publications 202 Acknowledgements 204 About the author 213


CHAPTER General introduction 1

Chapter 1 8 1. GENERAL INTRODUCTION 1.1 Why Why this study? Why for me? 1.2 When 1.3 Where 1.4 Who 1.5 What What test do we use? MoCA, CANE What do we mean (definitions)? Need, MCI, dementia, cognitive domains, depression, bipolar disorder, schizophrenia, doctors delay, patients delay 1.6 How Aim and outline of the dissertation

General introduction 9 1 1.1 Why 1.1.1 Why this study? In old age psychiatry, one can encounter a broad variety of referrals, at least in the Netherlands. Predominantly patients with affective disorders, anxiety disorders, psychotic disorders, and cognitive disorders are referred. Very few of the symptoms of these disorders are specific to only one psychiatric disorder and even ‘typical’ complaints can mimic, different aetiologies (American Psychiatric Association, 2013). This increases with age, as patients tend to have complaints in more than one specific domain, and often the complaints cannot be attributed to just one cause (Bierman et al., 2007; Schouws et al., 2012; Baune and Renger, 2014; Bora and Pantelis, 2015). Giving up hobbies is often wrongly attributed to being too old or being socially isolated, but can (also) be due to an affective disorder, negative symptoms of (late-onset) schizophrenia, side effects of medication (ranging from a tremor to cognitive impairment), or a developing neurodegenerative disorder expressing itself in apathy. Therefore, in old age psychiatry, complaints can result from more than one of the ‘classical psychiatric diseases’, as encountered in textbooks (Ferri et al., 2005). Often, age, frailty, social isolation, mobility, polypharmacy, comorbidity, and neurodegenerative diseases contribute to the overall picture owing to its population. This is most prominent in cases of cognitive impairment. A simple but daily example is a patient of age that has limited mobility, is depressed, uses psychotropic medication, and experiences (subjective) cognitive decline, and (therefore) quits his bridge club, for example, because of the shame of not being able to play at their former level. Giving up this activity can find its origin in either of these causes. Of course, none of these factors could be significant enough to cause mild cognitive impairment (MCI) by itself, but when combined, they would be. Furthermore, it can of course be neither of the above but an emerging neurodegenerative process or social deprivation causing depressive symptoms with subjective cognitive impairment causing fear of embarrassing oneself. Disentangling the possible aetiologies seems simple in theory, but the clinical reality is harsh. Simply stopping the antidepressant that seemed effective, to see if it was indeed the presumed side effects that caused cognitive impairment, is easier said than done. Alternatively, could it be that depression was only partly in remission, and cognitive impairment is a remaining symptom? Waiting for depression to subside does not take into account that cognitive deficits can linger even after the clinical depression has subsided (Ahern and Semkovska, 2017; Riddle et al., 2017; Semkovska et al., 2019). Was there already a neurodegenerative process developing in the background as 15% of 70 years and older and increasing to over 30% of 85 years and older have dementia, and

Chapter 1 10 even more so will have (mild) cognitive impairment (Volksgezondheidenzorg.info, 2019; ‘2020 Alzheimer’s disease facts and figures’, 2020)? Alternatively, were fear, shame, and loneliness key? Moreover, could disproving the subjective cognitive impairment through an objective test and organising transportation to the bridge club do the trick? For these questions to be answered with more certainty, one needs to judge cognitive complaints. Unfortunately, subjective cognitive complaints are poorly correlated with objective cognitive deficit (Pendlebury et al., 2015). This also accounts for next of kin reports and even more so for retrospective recall (Ryu et al., 2020). In addition, for being able to define a state, one must be able to compare it. This is often done in comparison to normative data, that is, the data of others resembling the patient in age and education. However, it would be even better to know the course and, therefore, be able to compare the patient with herself over time. In the above example, the issue would be simplified if there were a baseline at our disposal: was there already cognitive impairment, and if so, did the cognitive impairment change over time? During the depression, or was there a time correlation with recovery or after starting medication? There are many reasons why one wishes for some solid ground when assessing cognitive function in old age psychiatry. But the above all comes down to: Testing is objectifying. However, validated comprehensive tests that can help distinguish between the different aetiologies are not widely available in the short term. Therefore, not only is a short, rapid implementable test needed, but it also needs to be validated to interpret the results. Whether one needs to determine to exclude cognitive impairment or notice cognitive impairment at an earlier stage. 1.1.2 Why for Me? The Minimal Mental State Examination (MMSE) (Folstein, Folstein and McHugh, 1975) — I presume that most medical doctors are familiar with it — is a cognitive test to objectify cognitive impairment or screen for dementia or other severe cognitive impairments. It is a practical bedside test (a short questionnaire using only paper and pencil). Even though it was introduced in 1975, it still was at the beginning of my research, or even still is a sort of standard of short cognitive screeners when cognitive impairment is suspected (alzheimer-nederland.nl). However, its major shortcoming is a ceiling effect, meaning that less severe impairments would not be noticed by the MMSE or higher educated persons could pass the test even if they have impairments (Mitchell, 2017; Pinto et al., 2019). Therefore, at the beginning of my career as an old age psychiatrist, I was trying to select a cognitive test that would better fit my daily clinical use, including house visits, in correctly

General introduction 11 1 identifying mild cognitive impairments. Therefore, they still need to be easily applicable, accessible, and affordable. During a home visit consultation in 2008 regarding a patient with lithium intoxication, I also added routine cognitive screening. In this case a ‘new’ cognitive screener: the Montreal Cognitive Assessment (MoCA) was used (Nasreddine et al., 2005). This was more to gain experience with new cognitive screening instruments and which one to use from now on as I did not expected major abnormalities. This is partly because I had gone through an extensive complaint anamnesis with the patient, which was focused on her lithium use and its side effects. As it turned out, I overestimated that significant cognitive abnormalities would have been ‘à vue’ for me as an old age psychiatrist to notice. You are holding the result of that test in your hands in the form of this dissertation. Of course, this example could not be the only patient with unknown cognitive impairment, intervening with the treatment. In her case, there was a (lithium) intoxication as she forgot that she had already taken her lithium and therefore mostly likely doubled her dose by mistake. Not considering this would be a mistake and prone to more foreseeable mistakes. Being confronted withmy own shortcomings, even though the reason for the consultation had a different origin, I had fallen into a classical nicely put ‘doctors delay’ trap. I realised that the needs patients mention do not necessarily correspond with the needs the doctor hears, sees, or thinks should be met (by him or her). Furthermore, even those needs that the patients want or need to be met are not necessarily the ones they mention to be met, that is, ‘patients delay’. This can have multiple reasons, ranging from being unaware of their need to denial of their need, fear of being stigmatised, or thinking a doctor cannot help them with their particular needs. These “delays” result in the following question: When should we screen? 1.2 When The discussion above illustrates well the dilemma that doctors face on a daily basis. To what extent do they have to look for something that is not (yet) a problem, that is not seen or experienced as such (by the doctor or by the patient), or is not recognised as such. Or, when it is a problem but not (yet expressed as) a complaint? Does the patient always have the right to downplay or even ignore the problem? The COVID-19 pandemic is illustrative in that there is no obvious answer on second sight. This dilemma occurs also often with cognitive impairment, as many patients don’t want to have an elaborate

Chapter 1 12 cognitive assessment as they find it a hassle, too demanding or deem it ‘much ado about nothing’. Sometimes, they do not want to be confronted or diagnosed with cognitive impairment. This can be due to multiple reasons, ranging from fear of losing their driving licence to denial of the impairment. Another situation is that the patient is not aware of their cognitive decline and only afterwards realises the consequences of a diagnosis. However, this right to ignore can result in dangerous situations, such as cooking at home resulting in the cooking stove not being turned off afterwards. Is this a problem for the doctor to resolve? Is it only his or her concern or responsibility if a complaint or problem leads to a medical problem? Doctors primarily want to help. However, the Hippocratic Oath states “First, do no harm” (Latin: Primum non nocere). To what extent is helping the individual lead to harming others, for example, by screening all patients to find a few cases and using scarce resources? Or ‘helping’, that is, not diagnosing a patient on their request so they won’t be stigmatised. Therefore, avoiding a ban on driving but perhaps becoming a risk for all traffic participants? This is plenty of material to consider and debate. On many levels, medical, ethical, philosophical, political, and so on, all the way up to the (patients) kitchen table. Therefore, this debate is far from over. The debate varies across subjects and settings, ranging from clinical themes (e.g. diseases) to social (e.g. loneliness) or financial (e.g. healthy food) issues. One must bear in mind that the outcome of the debate will be different for screening individual patients than for screening the general population. In clinical practice, doctors often use guidelines that give them something to hold on to. One of these ‘guidelines’ is a list of requirements, or rather criteria, for screening populations, which was drawn up in 1968 on behalf of the World Health Organization WHO (Wilson and Jungner, 1968). They were still relevant up until today (Sturdy et al., 2020). These criteria are summarized by the RIVM (Dutch national health institute) as follows: ‘....a screening that falls under the national population screening programmust be of benefit to participants, voluntary, and scientifically based. To determine whether a screening is justified, international criteria were drawn up by Wilson and Jungner in 1968’.

General introduction 13 1 Table 1. Criteria of Wilson and Jungner (1968) 1 The disease to be detected must be a major health problem. 2 There must be a generally accepted method of treatment for the disease. 3 There must be adequate facilities for diagnosis and treatment. 4 There must be a recognizable latent or early symptomatic stage of the disease. 5 A reliable detection method must exist. 6 The detection method must be acceptable to the public. 7 The natural course of the disease to be detected must be known. 8 There must be agreement as to who should be treated. 9 The cost of detection, diagnosis and treatment must be in an acceptable proportion to the cost of health care as a whole. 10 The process of detection must be a continuous process and not a one-time project. In 2008, a list of additional criteria was drawn up by the World Health Organization (WHO). Although the abovementioned criteria are intended for large-scale screenings, such as national population studies, they can also be used as guidance for small-scale screenings. Think of local initiatives or specific patient groups. It becomes less clear when screening is used on an individual basis for complaints that would otherwise be overlooked. Or when it concerns using the screener as a severity scale. It is doubtful whether this is still a screening in itself, or whether it is more about using a screener to follow the course of a disease for an individual situation. Other ethical criteria, which are usually enshrined in health laws, will come into effect. However, what if the “on indication” is applied to all (referred) patients? Therefore, we will consider the WHO criteria as a starting point, but not as a rule. Several criteria enumerated by the WHO are touched upon in this dissertation, whereas many are not answered for various reasons. The most important reason is that the criteria themselves are not the subject of our study, but we use them as guidelines. As the above personal clinical experience illustrates, it is easy to overlook (or ignore) other matters that are at hand, such as social isolation or cognitive impairment (the underlying cause of a problem?), when your attention is focused on something else (the main complaint). In particular, when efficiency is expected due to time constraints and costs, to minimise waiting lists. Screening can never be a substitute for a thorough diagnostic workup. However, can screening not be a helping hand? If so, when should one consider screening? In addition to the above criteria, when a screening is justified, there are also costs. These costs should not be limited to the financial sphere (Table 1, criteria 9). There are many

Chapter 1 14 factors to be considered besides spending on resources, efficiency, and so on. Screening takes its toll in many areas, but the benefits of screening should always outweigh its disadvantages. These negative factors also depend on where, that is, in which setting or for which population, the screening takes place. Haphazardly screening the general population costs more than it will yield. Therefore, it should not only be considered when (or for what) to screen but also where to screen. 1.3 Where Particularly in geriatric psychiatry, the symptoms of several aetiological entities resemble or even overlap each other. Not only is it difficult to distinguish the aetiologies one from the other, but sometimes they coexist and contribute to the same complaint (to some extent). One particular entity is always present in an old age psychiatric practice: old age. Where living its life has left its mark on many patients. However, where do these traces of old age turn into ‘no longer appropriate for age, education, and social context’? Alternatively, they have a (negative) impact on the quality of life, even if they are (still partly) appropriate for their age or social context. When is an intervention justified if they negatively affect ‘wellbeing’? How much of it must deteriorate before it can be called a disease? Who is to judge if someone’s ‘wellbeing’ needs to be improved? If so, which domains should be prioritised? Is that for the doctor to decide, the next of kin representing society, or is the decision of themain character, that is, the patient? This accounts for different domains, such as social, psychological, and physical, but this is especially true for cognitive impairment. These questions or dilemmas can only be solved if one also has insight into the complaints and how much they play a role: the quantitative part. In old age psychiatry, these factors play a significant role as they tend to add up known as ‘frailty’, more than in other disciplines where they may seem to be in the background or have no influence at all. These factors not only include age-appropriate problems, but also the field of geriatric psychiatry. Geronto-psychiatry and psychogeriatrics have always been the core of old age psychiatry. Where the psychiatrist for (younger) adults can focus on the complaints of her patient without worrying about age-related complications or comorbidities, in old age psychiatry this will always have to be considered. However, the extent to which age alone plays a role in the clinical picture of elderly patients, and to what extent (complications of) age cause a different form of the disease or even another disease? Is it more of the same problem (compared to a younger patient) or is

General introduction 15 1 it a different problem? There are indications that in older patients with, for example, a bipolar or depressive disorder, a different manifestation or even a type of this disorder may be at hand (Sajatovic et al., 2015; Aizenstein et al., 2016). In particular, if the disorder arises only for the first time at a later age (i.e. late onset or very late onset), the expected neurodegenerative disorders that occur with advancing age can also play a role, in both numbers (prevalence) and severity. One has to bear in mind that the prevalence of dementia is already up to 10% from 65 years of age and above (Volksgezondheidenzorg. info, 2019; ‘2020 Alzheimer’s disease facts and figures’, 2020). Our colleagues, who also deal with neurodegenerative disorders (neurologists, clinical geriatricians), normally do not often have other psychiatric diseases among their referrals that cause cognitive impairment. Of course, they still need to be aware of them. In The Netherlands, referrals to old age psychiatry consist of a mix of neurodegenerative and other psychiatric disorders, such as depression, bipolar disorders, schizophrenia, and severe anxiety disorders, all of which can be accompanied by poor (long-term) cognitive functioning (Bierman et al., 2005; Schouws et al., 2012; Baune and Renger, 2014; Bora and Pantelis, 2015; Ahern and Semkovska, 2017; Riddle et al., 2017; Semkovska et al., 2019; Van Rheenen et al., 2019). In contrast, dementia can frequently be accompanied by depression, hallucinations, delusions, and anxiety (Lyketsos et al., 2002; Di Iulio et al., 2010). Both neurodegenerative and psychiatric diseases often present themselves with symptoms normally attributed to the other entity before their ‘classic features’ appear (Lyketsos et al., 2002; Reichenberg, 2010; Eikelboom et al., 2021). We will later elaborate on the overlapping presentation in the definition paragraph. However, for these questions to be answered, one needs insight into the causal entities’ or aetiology behind the complaint: the qualitative part. In addition to the ‘quantitative’ and ‘qualitative’ reasons for knowing the cognitive function of a patient in old age psychiatry, there are other reasons for cognitive screening. As mentioned above, the prevalence of MCI and dementia above 60 years of age is high in the general population, and this will be even higher in an older psychiatric setting. The population of older people is increasing owing to demographic factors. This eventually results in more older patients having psychiatric problems. This, in turn, will increase the number of psychiatric patients with cognitive complaints in addition to the expected increase in patients with neurodegenerative disorders. This will lead to an increase in referrals to older psychiatric clinics for patients with cognitive impairment. Together with more awareness due to public campaigns on cognitive impairment, there is also a trend of being assessed earlier in the process with fewer complaints (Grimmer et al., 2015). Some of themhad only subjective complaints without being able to objectify these complaints. Even

Chapter 1 16 so, most patients with dementia will stay undiagnosed, 50% in developed countries and up to 90% in poor countries (Alzheimer’s disease International, 2016). A part of the public campaigns by advocacy groups or policymakers is to stimulatepatients aswell as healthcare professionals for more (early) diagnoses. This will result not only in more referrals but also in more referrals with less well-described cognitive impairments, and it will be harder to differentiate aetiologies (Mitchell, 2009). A detailed neurocognitive assessment, which is costly, time-consuming, and not widely available, is advised by advocate groups in the case of cognitive complaints. However, doing so for all patients with cognitive complaints will be an assault on available resources. The cognitive diagnostic tracks in memory clinics or old age psychiatry clinics are already being challenged and will be further challenged due to the increase in the older population (Alzheimer’s Disease International, 2018). Moreover, many subjective or even objective (mild) cognitive complaints (up to 40%) will subside or decrease over time (Alexopoulos et al., 2006). Selecting patients who are in need of this elaborate neurocognitive assessment would help reduce the burden on resources. We have to find patients who benefit from an elaborate neurocognitive assessment better, or in other words, triaging those who are in need of a specialised neurocognitive assessment and who are not (yet) in need of such assessment. Unfortunately, this leads to more questions: ‘who’ is in need and ‘how’ do we find them? 1.4 Who A way to achieve early detection and to find those in need of an elaborate assessment is through screening. Advocacy groups for dementia encourage this, but the debate on whether screening is a solution or wise is still debatable (Borson et al., 2013; Davis et al., 2015; Burn et al., 2018). Many factors must be considered, such as spending resources and efficiency. One of the main issues is which population to screen, for what purpose, and what instrument to use (Janssen et al., 2017). As screening for cognitive impairment in a general practitioner’s office will result in different findings than at a memory clinic or in old age psychiatry, more people are likely to have cognitive impairment. Screening with a fast and cheap instrument will result in different findings than screening with a more time-consuming and multi-factor full assessment. The first example yields a high quantity with low quality, but the latter example, on the contrary, yields high quality and low quantity. Additionally, the purpose of screening should be considered. In a test, specificity and sensitivity always compete with precedence. Which one should prioritise depends on the purpose of the test. Screening for HIV can serve as an example of this. At the doctor’s office, you want to establish a definite diagnosis. You do not want to

General introduction 17 1 diagnose someone falsely. This translates into no ‘false positives’, that is, high specificity. However, at the blood bank, you want to rule out the disease with certainty. This, in turn, translates into no ‘false negatives’, you do not want to miss a case (high sensitivity). In old age psychiatry, regarding cognitive complaints, we can organise the population by cognitive complaints, cognitive functioning, and whether they have psychiatric complaints or not. Creating groups using these three parameters results in: not suspected of cognitive impairment (but with psychiatric symptoms), suspected of but not objectified (with or without psychiatric symptoms), and objectified cognitive impairment (with or without psychiatric symptoms). Figure 1. Levels of certainty of cognitive impairment in old age psychiatry and who to screen. This results in different levels of selection for screening: screening all referred patients to old age psychiatry, or only those considered at high risk. Screening all referred patients during an initial interview brings benefits in the form of knowing patients’ cognitive status, besides avoiding doctor delays or patient delays. These benefits go beyond just (early) detection of cognitive impairment present at the time of the initial history interview that would otherwise be unnoticed. They can also help with issues in the near future such as foreseeable cognitive problems due to prevalence, psychotropic medication, or psychiatric episodes, among others. Of course, screening patients without complaints and at lower risk is not without a downside. The most prominent are the financial cost, psychological burden, and false positives, but all Wilson and Jungner’s criteria apply. On the ethical side is the unexpected discovery of cognitive impairment with major social consequences for patients. The more the population that is screened is preselected

Chapter 1 18 and is at an increased risk of cognitive impairment, the less this downside will be an issue. However, this preselection of the screened population comes with missing the benefits of screening all, next to the chance of missing unseen cognitive impairment. In short, there are advantages and disadvantages of a screening. They depend on why, when, where, and who is being screened. And of course ‘With What’ are we going to screen is of significance, too. 1.5. What 1.5.1 What test do we use? Cognitive complaints are a core feature of many (unmet) problems encountered in old age psychiatry and often remain hidden, deliberately or not, but have a major impact on the treatment and quality of life. Therefore we want or even need to know the cognitive state during the initial history interview of referred patients. In 2008, I sought a substitute for the MMSE because it lacks the ability to detect MCI (Pinto et al., 2019). As explained above, in old age psychiatry, we encounter a lot of subjective cognitive impairment due to various possible aetiologies that are not detectable by the MMSE. Therefore, this engenders the problem of not being able to objectify these complaints and follow their course. Were they too subtle to be detected or were the complaints only subjective? It is known that the MMSE cannot detect mild impairment well, as it is not designed to detect MCI. It was designed as a short and fast test for detecting major cognitive impairment. Part of the MMSE, a memory test, asks for three words to be remembered: the MoCA asks for five words. However, during an elaborate neurocognitive assessment, participants are asked to remember 15 words. The latter takes considerably more time, but no ceiling effect occurs, as the median number of words to remember is approximately six to seven. In addition, a learning effect can be observed, since the 15-word test is repeated more often. This example illustrates well the tension between trying to be fast and trying to be complete, limiting wrong conclusions. A bedside test that could objectively assess the cognitive state, including MCI, of our patients quickly, cheaply, conveniently, and sensitively enough to detect mild cognitive impairment, was needed. Among others, I considered the 2-minute test, 7-minute test, clock drawing test, ACE/ACE-R (Addenbrooke’s Cognitive Examination), and the CAMCOG (in part). Because of an educated guess, I chose the newly introduced MoCA.

General introduction 19 1 Montreal Cognitive Assessment ‘MoCA’: The MoCA is a widely used short screening tool for MCI and mild dementia (MD). It was introduced and validated in French and English in 2005 (Nasreddine et al., 2005). Until now, it has been validated in multiple settings and languages, although not in psychiatry (mocatest.org). It is now recommended by several institutions and guidelines, including Cochrane and Alzheimer International, to use it as a screener for cognitive impairment (Davis et al., 2013; Alzheimer’s disease International, 2016). Its use and popularity are growing fast, and it seems to be rivalling the MMSE. There are 867 publications to date (d.d. January 2022) with the MoCA as the main subject (mentioned in the title) and even several more using the MoCA in their study to measure cognition (9722 with MoCA as a keyword with Embase). The MoCA consists of one page covering the cognitive domains of executive function and visuospatial abilities, naming, short-term memory, attention and working memory, language, concentration, verbal abstraction, and orientation. It can be performed within 10 minutes, with a maximum score of 30, indicating that no errors were made. Scores can be corrected for low education according to instructions by adding one point to the total score of patients with 12 years of education or less. Three validated versions differ from each other in minor ways to avoid a learning curve. The nature or subject of the questions remained the same, but the numbers or words differed between versions. For example, version one asks to subtract 7 from 100, whereas version 2 asks to subtract 6 from 100. The originally suggested cut-off for the diagnosis of cognitive impairment was a score of < 26 (less than 26). In the original study, the MoCA was compared to the MMSE. The results showed that it was superior to the MMSE in detecting MCI and mild AD. At a cut-off <26, the sensitivity of the MoCA was 90% for MCI and 100% for mild AD. This was 18% and 78% for the MMSE, respectively. The MoCA’s specificity was 87% compared to 100% for the MMSE. Indicating that the MoCA was too difficult for 13% of the nonimpaired, but the MMSE was too easy for 82% of the mildly impaired and 12% of the people with mild dementia. However, the results of the Dutch version in patients with cognitive symptoms in a geriatrics department deviated from this for unknown reasons, with a sensitivity and specificity of 72% and 73% for MCI, respectively, compared to healthy controls (Thissen et al., 2010). Validation has also been performed in several specific populations, including vascular dementia (Ihara et al., 2013), frontotemporal dementia (Freitas et al., 2012) and Alzheimer’s disease (Freitas et al., 2013).

Chapter 1 20 Depending on the population in which the test is administered, reliability changes. The positive predictive value (PPV) decreases and the negative predictive value (NPV) increases when there are fewer cognitive impaired patients in the studied population. For example, for the MSSE in a memory clinic, the PPV is 86% and NPV is 73%, and in general practice, the PPV is 54% and NPV is 96% (Mitchell, 2009). A test should be validated for specific populations to maintain high reliability (Rossetti et al., 2011). The reliability changes are partly due to the prevalence of the target disease in the population. However, as explained before, in old age psychiatry, the symptoms overlapmore between the diseases and, therefore, it becomes more challenging for a test to identify the target condition. It should be noted that the MoCA tests a state and not a disease. Theoriginal validation study (Nasreddine et al., 2005) uses healthy controls for comparison. Although this is often done in validation studies, it is prone to introduce bias (Davis et al., 2013; Bossuyt et al., 2015). This will especially affect the specificity, as the comparison group will have clinically unrealistically high (good) MoCA scores. Healthy controls would normally not reach out for an assessment, as they are selected to have no cognitive complaints, impairments, or any other disease that could cause cognitive complaints. Therefore, separation from the impaired group will be too optimistic. This results in unnaturalistic high specificity. In clinical practice, an assessment should identify impaired patients in a naturalistic population, who, in our case, are patients that are referred (with complaints). If the MoCA is used for screening purposes, the setting (population) is of great importance. Healthy controls are seldom included in the target population. There is since the MoCA was introduced more literature on the effects of the (study) population on cutoffs. Normative data are also available. In short, multiple studies have suggested that the original cutoff creates too many false positives, and the specificity is expected to be lower in a clinical setting (Davis et al., 2015; Carson, Leach and Murphy, 2018; Elkana et al., 2020). This was partly expected as it used healthy controls, but other parameters that influence MoCA scores have emerged besides education. The most prominent are age and social status (Pinto et al., 2018). Therefore, it is clear that the MoCA, as with other tests, should be validated in the setting and population where they are going to be used. In our case, one can debate whether this is for all referred patients to old age psychiatry or only for those suspected of having cognitive impairment.

General introduction 21 1 Camberwell Assessment of Need for the Elderly ‘CANE’: For finding the ‘unmet needs’ and ‘needs’ we used the CANE (Camberwell Assessment of Need for the Elderly) which has been validated already (Reynolds et al., 2000). The CANE is a semi-structured interview based on the Camberwell Assessment of Need (CAN) (Phelan et al., 1995), which is adapted for the elderly. It was developed to ‘measure the needs of people in the general adult population with severe mental illness. It is based on a model of need as a subjective concept, accepting that there may be differing but equally valid ideas about the existence of a need’. Therefore, the needs are not solemnly scored from the professionals’ perspective, but also from the experience of the patient and (especially important for patients with cognitive impairment such as dementia) from the caretakers’ perspective. This results in three different scores from three different points of view that can be compared. Where ‘Identifying a need means identifying a problem plus an appropriate intervention which will help or alleviate the need’ or ‘a need was thought to be present when a patient’s level of functioning falls below or was threatened to fall below, some minimum specified level and if a potentially effective remedy existed’. It consists of 24 questions covering 24 areas divided into 4 domains (environmental, physical, psychological, and social needs). Each of the 24 areas or ‘items’ is scored on a 3-point scale. This item or need can either be: no problem, that is, ‘no need’ (0 points); no/moderate problem because of continuing intervention, that is, ‘met need’ (1 point); and current serious problem, irrespective of any on-going intervention, that is, ‘unmet need’ (2 points). The duration of administration to the patient, next of kin, or clinician are 30, 20, and 10 minutes respectively. 1.5.2 What do we mean (definitions)? The abovementioned personal clinical experience reveals multiple uncertainties. These uncertainties led to questions. We aim to reduce uncertainty of some (by no means all) of these questions by our study. However, before addressing these uncertainties, definitions must be introduced to avoid further confusion. ‘Need’: In the paragraph above, the word ‘need’ was introduced. In general, a ‘need’ can be translated to, amongst others, require (something) because it is essential or very important rather than just desirable (Google.com, 2021) or the things that a person must have in order to have a satisfactory life (Cambridge dictionary). ‘Need’ in the context of well-being probably

Chapter 1 22 reminds most readers of the Maslow pyramid (Maslow, 1954, 1970). In our study we use the definition of – and objectively define ‘needs’ by – the Camberwell Assessment of Needs for the Elderly (CANE). The CANE incorporates the Maslow findings as well as the taxonomy proposed by Bradshaw (Bradshaw, 1972) involving normative needs (by experts), felt needs (by patients), expressed needs (or demanded), and comparative needs (with other patients) from a sociological perspective. The aforementioned contemplations or dilemmas I experienced (in the Why paragraph) is well-described by the concepts presented in the dissertation of Reynolds (Reynolds, 2003) on the CANE that are: Need: What people benefit from; Demand: What people ask for; Supply: What is provided. How these three concepts are interpreted is prone to change over time for numerous reasons, but most noticeable to knowledge (of science, e.g. doctor, as well as the patients/public) and resources (in money as well as technics). The results of these three interpretations will have an effect on doctors’ responses to the dilemma mentioned above, that is, when should a doctor (re)act even if a need is not expressed as such? This includes the question of whether screening is desirable. It is, therefore, not surprising that these three concepts (need, demand, and supply) are, to some extent, reflected in Wilson and Jungner’s criteria listed above, although the criteria in themselves did not change over time. One could say that the questions remain the same, but the answers will change. The DSM IV andDSM5 list the diagnostic criteria for different psychiatric and neurocognitive diseases (NCD). Without trying to be complete but also trying not to copy the DSM, we will summarise and capture its essence here as an introduction with an emphasis on cognitive impairments. ‘Mild cognitive impairment’: Mild cognitive impairment, often abbreviated as MCI, was (re)introduced by the Mayo clinic in 1999 and 2004 (Petersen, 2004). The definition of MCI has evolved over the years; however, the idea behind it remains the same (Table 2). The idea behind introducing MCI is that cognitive impairment is a state on a continuum ranging from normal cognition on one side to dementia on the other end of the continuum. ‘It should be considered as a description of cognitive functioning in which the underlying disorder can vary rather than as a nosological entity representing the prodromal stage of AD’ (Visser and Verhey, 2008). However, the original starting point of the concept focused on Alzheimer’s disease (AD); therefore, the theory was more towards MCI being a state before the transition towards Alzheimer’s disease. In recent decades, there has been a shift in awareness that, even though Alzheimer’s disease is the most frequent cause of

General introduction 23 1 dementia, dementia has multiple aetiologies. With this shift, the concept of MCI has also changed. From a prodromal or predementia concept of ‘… a transitional period between normal ageing and the diagnosis of clinically probable very early AD, and this transitional zone has been described using a variety of terms such as mild cognitive impairment’ (Petersen, 2004) towards a more descriptive intermediate functional state to ‘define the grey area between intact cognitive functioning and clinical dementia’ (Petersen et al., 2014). With the emancipation of other aetiologies besides AD that can cause cognitive impairment, the criteria of MCI were also adapted. From primarily memory-focused criteria towards ‘a condition in which individuals demonstrate cognitive impairment with minimal impairment of instrumental activities of daily living’ (Petersen et al., 2018). Most importantly ‘it can also be secondary to other disease processes i.e., other neurologic, neurodegenerative, systemic, or psychiatric disorders’ (Petersen et al., 2018). With this shift, both cognitive and functional abilities must be considered in the evaluation of MCI (Winblad et al., 2004). In turn, these adaptations initiated further specification of the MCI. This involves differentiating between amnestic (aMCI) or non-amnestic (naMCI) disorder and with or without multiple domains. This results in four subtypes: aMCI (single or multiple domains), naMCI (single or multiple domains) (Petersen, 2004). Although MCI is less AD-focused, AD is still a frequently suspected probable cause. This is reflected in new subtypes like ‘MCI due to AD’ or ‘MCI supported with biomarkers’, although they are mainly created for research purposes (on early interventions for AD) (Mattsson et al., 2009). One reason for this is that not all patients with MCI develop dementia. Even more, there is substantial literature that 20% (up to 55% are reported) revert to normal cognition, 40% remain stable, and 40% convert to dementia. By creating subgroups of MCI, studies attempt to predict who is at a higher risk of developing dementia (Visser and Verhey, 2008). Although the percentages of reverting and converting differ substantially in the literature, it is acknowledged that people with MCI have a significantly higher risk of progressing to dementia than agematched controls (Alexopoulos et al., 2006; Petersen et al., 2018). This difference in conversion rate as well as the prevalence and incidence that vary across studies is most likely due to the differences in study populations, as incidence increases with age and the type of patient (Visser and Verhey, 2008; Bermejo-Pareja et al., 2021) — 5% to 15% for the annual conversion rate to dementia compared to 1–2% for controls, resulting in a five to ten times higher conversion rate. Cumulative dementia incidence was 14.9% in individuals with MCI older than age 65 years, during following 2 years. MCI prevalence was 6.7% for ages 60–64, 8.4% for 65–69, 10.1% for 70–74, 14.8% for 75–79, and 25.2% for 80–84 (Petersen et al., 2018).

Chapter 1 24 Table 2. Criteria for (M)CI over the years. Original 1999 Mild Cognitive Impairment Criteria Recommendations General criteria for MCI 2004 Minor NCD DSM 5 2013 Major NCD DSM 5 2013 -Memory complaint, preferably corroborated by an informant -Memory impairment documented according to appropriate reference values -Essentially normal performance in nonmemory cognitive domains -Generally preserved activities Not normal, not demented (Does not meet criteria (DSM IV, ICD 10) for a dementia syndrome) Cognitive decline: -Self and/or informant report and impairment on objective cognitive tasks -Evidence of decline over time on objective cognitive tasks and / or Preserved basic activities of daily living / minimal impairment in complex instrumental functions Moderate Cognitive Decline • NOT Interfere with independence • Not due to delirium • Not due to other mental disorder Significant Cognitive Decline • (minimal) interfere with independence in everyday activities (ADL) • requiring assistance with instrumental activities of daily living (IADL) • Not due to delirium • Not due to other mental disorder ADL; activities of daily living. IADL; instrumental activities of daily living. NCD; neurocognitive disorder. ‘Dementia’: I often tell my patients that dementia is not a disease but an agreement between doctors. It defines a cognitive state that can be caused bymany different aetiologies, andmore than 50 cases have been reported. However, a large majority of these causes are attributed to Alzheimer’s disease (AD). Other common causes are vascular dementia, Lewy body dementia, and frontotemporal dementia. As always, we (the doctors) try to simplify things for our patients but not ourselves. Thus, there are many different agreements on what is considered dementia, but these agreements differ from each other. As our study was based on psychiatry, we used the DSM-IV, and later, the DSM 5 classification (the fourth and fifth editions of the Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association, 2000, 2013)) in the lead for dementia. However, we also incorporate the different classifications of different specialist or advocate groups per clinical disease, for example, NIA-AA/NINCDS-ADRDA for AD (McKhann et al., 2011). The WHO describes dementia as ‘a syndrome occurring as a result of disease of the brain, which is usually chronic or progressive innature. It consists of impairment of several higher cortical functions, which include memory, thinking, comprehension, calculation, learning, language and judgement. These impairments often occur alongside changes in emotional control, social behaviour or motivation. Alzheimer’s disease and cerebrovascular disease are among the causes of dementia’. In general, one could summarise the global concept of ‘all-cause’ dementia as ‘the cognitive deficits are sufficient to interfere with independence

General introduction 25 1 and show a decline (from a previous level), that is, requiring (minimal) assistance with instrumental activities of daily living (IADL)’. These definitions attempt to create a theoretical yes or no situation or a sharp line on the cognitive continuum. However, in clinical practice, there is a grey area, or rather a rainbow, of interpretation differences. Attempts have been made to categorise this continuum. The DSM 5 uses the words ‘modest’ versus ‘substantial cognitive decline from a previous level of performance in one or more of the domains’ to capture the difference between not (yet) having dementia (minor Neurocognitive disorder (NCD)/MCI) and patients with dementia in words. An attempt to operationalize this is by stating ‘test performance in the range of one and two standard deviations below appropriate norms’ for minor versus ‘test performance in the range of two or more standard deviations below appropriate norms’ for major NCD. This translates to a score between the 3rd and 16th percentiles for minor NCD and below the 3rd percentile for major NCD, whereas the (amnestic) MCI was defined to have a delayed recall of 1.5 standard deviations below appropriate norms on a 15-word verbal learning test (Petersen et al., 1999). One method is to quantify the cognitive continuum by using severity or rating scales. The CDR (0-3) (Hughes et al., 1982) and GDS (1-7) (Reisberg et al., 1982) are the most wellknown (Table 3). This is to (try to) objectify the staging of cognitive impairment/dementia and is used for multiple purposes such as research, renewal of the driving licence, and nursing home allocations. Table 3. Rating scales for cognitive impairment. CDR Clinical Dementia Rating Scale GDS Global Deterioration Scale CDR0 No cognitive impairment GDS1 No cognitive impairment GDS2 Age-associated impairment CDR0.5* Very Mild Dementia GDS3 MCI CDR1 Mild Dementia GDS4 Mild Dementia CDR2 Moderate Dementia GDS5 Moderate Dementia CDR3 Severe Dementia GDS6 Moderate Severe Dementia GDS7 Severe Dementia * In clinical practice, CDR0.5 is often considered as equivalent to MCI but formally, it is already called dementia while this is an exclusion criterion for MCI Another factor in dementia diagnostics is to express the diagnostic certainty in terms of probability. This practice of the Alzheimer diagnostic guidelines of the NIA-AA/NINCDSADRDA (McKhann et al., 2011) are spreading to other diagnostic guidelines. The additions

Chapter 1 26 ‘possible’ and ‘probable’ are most often used to quantify the amount of and/or core criteria met with respect to the classification guidelines with ‘unlikely’ and ‘definite’ (e.g., evidence of AD via autopsy or biopsy) on either side. These additions also represent well the specificity or uncertainty that these classifications still obtain. Psychiatric symptoms in dementia, often expressed as behavioural and psychological symptoms of dementia (BPSD), are not only important in invalidating symptoms and lowering the quality of life, but they can also have a diagnostic and predictive role (Defrancesco et al., 2020). ‘Affective syndromes characterized by depressive symptoms are associated with faster functional decline whereas Manic syndromes are better at predicting cognitive decline’ (Palmer et al., 2011). However, as up to 50% and 80% of the patients with MCI and dementia, respectively, exhibited (relevant) (neuro)psychiatric symptoms from the month of onset of the cognitive symptoms or the month prior to the diagnosis, these symptoms can not only have a predictive role but also mimic psychiatric diseases and frustrate the diagnostic process (Lyketsos et al., 2002; Eikelboom et al., 2021). ‘Cognitive domains’: Cognitive impairment may occur in different forms or functions. These functions can be categorised into different domains. Again, how or where the functions are categorised can differ, depending on the literature. Most often, we distinguish attention, planning, inhibition, learning, memory, language, visual perception, spatial skills, social skills, and other cognitive functions. In short, we follow the below (DSM-5) descriptions of the domains: complex attention, executive functions (attention, planning, inhibition), language, learning and memory, perceptual motor function (visual perception, spatial skills), and social cognition.