Assessment scales in dementia

Abstract

Dementia involves progressive and often remorseless decline in cognition, function, behaviour and care needs. Assessment in dementia relies on collateral as well as patient-derived information. Many assessment scales have been developed over decades for use in dementia research and care. These scales are used to reduce uncertainty in decision making, for example in screening for cognitive impairment, making diagnoses of dementia and monitoring change. Ideal scales used in dementia should demonstrate face validity and concurrent validity against gold standard assessments, should be reliable, practical, and should rely on objective rather than subjective information. Assessment scales in the domains of cognition, function, behaviour, quality of life, depression in dementia, carer burden and overall dementia severity are reviewed in this article. The practical use of these scales in clinical practice and in research is discussed.

Introduction

Dementia is a term for a clinical syndrome characterized by progressive acquired global impairments of cognitive skills and ability to function independently. Many patients show varying levels of behaviour disturbance at some point in the illness. Care burden, for family carers as well as state/other care funders, increases as the condition progresses. The syndrome is caused by many diseases, with Alzheimer’s disease, vascular dementia and dementia with Lewy bodies together accounting for around 90% of cases. Incidence and prevalence of dementia are strongly age dependent. With global aging of populations, dementia prevalence is rising and is projected to continue to do so for much of the present century. The collateral damage in dementia is vast. Carer burden in terms of physical work, psychological distress and financial obligations is great. Many nonspecialist branches of medicine now operate some system for screening for and diagnosing dementia – for example, primary care, neurology or general hospital inpatient services. Rating scales are often advocated for use in influential guidelines [NICE, 2006].

Assessment scales in dementia

A vast industry in generation, validation and reporting of properties/utility of rating scales in most branches of medicine, including dementia, has emerged. Many scales have been devised just in the field of dementia [Burns et al. 2002]. The purpose of an assessment scale is to increase the precision of a decision by reducing subjectivity and increasing objectivity; for example, using a cognitive screening test score to screen for underlying dementia, to distinguish impairment due to dementia from normal age-related cognitive change or to monitor the effects of treatment of dementia in a clinic or controlled trial. The properties of an ideal assessment scale would be that it is valid, that is, it has face validity (experts like clinicians, patients and carers would agree that the questions are relevant and important), that it has construct validity (it measures the construct it was designed to measure), concurrent validity (when used alongside a gold standard assessment like a very well validated scale or an expert clinical assessment, it performs well), that it shows reliability – typically inter-rater reliability (two or more raters using the scale in the same subjects and conditions come up with the same result) and test–retest reliability (the same rater using the scale on another occasion in the same subject comes up with the same result). Importantly, it should be practical to use – in practice, this often depends on it being short (so it can be used in busy clinical practice or as an outcome measure in a trial such that participants are not overburdened by long interviews) and acceptable – so it does not upset, exhaust or embarrass the patient or assessor. The key task in using assessment scales in dementia (as in any field) is clarifying what they are to be used for, and by whom. Scales are frequently misunderstood and misused, wasting patients’, carers’ and assessors’ time. Another aspect of dementia which distinguishes it from other progressive neurological disorders is the increased reliance on others to assess clinical and practical problems. Dementia may from its earliest stages affect judgement, speech and memory, making patient judgements less reliable. Proxies such as family or professional carers need to be consulted at all stages in the care journey, altering the traditional assessment method to a shared patient/carer encounter (for example, the combination of a patient-facing cognitive assessment with a structured or unstructured informant interview in diagnosing dementia). This is directly relevant to the choice of assessment scales to be used in dementia care and research. In particular, judgements about functional impairment, quality of life and behaviour problems may have to be mainly, or entirely, derived from proxy reports.

An overview of assessment scales in dementia

In clinical practice and in research, cognition is considered the key change we want to observe in people with dementia. Diagnostic criteria for dementia depend on the presence of cognitive impairment [APA, 2000], and other aspects of the clinical picture in dementia (behaviour, impairment in function, increased costs, carer stress) ultimately derive from impaired cognition. Function refers to abilities to carry out activities of daily living, a direct consideration at the point of diagnosis of dementia [APA, 2000] and also in assessing change and planning care interventions. Behaviour changes seen in dementia, often referred to as Behavioural and Psychological Symptoms in Dementia (BPSD) are of special importance in influencing prescribing (often hazardous), institutionalization of patients and carer stress. Proper evaluation of interventions for BPSD can only be carried out using reliable scales. Quality of life (QOL) is a multidimensional concept which reflects the patient’s perception of the effect of their illness on their everyday physical and emotional functioning. Measurement of QOL is increasingly popular. In dementia, subjective evaluations are frequently impossible, and patients and carers have very different ratings of QOL. Scales for measuring QOL include patient and proxy versions, and generic and dementia-specific scales. Depression is common in dementia; rating this fundamentally subjective experience is especially challenging in patients with cognitive impairment. Carer burden is a major issue in dementia; service- and research-level interventions may look to measure effects on carers using generic measures of psychological distress or measures designed specifically to measure carer burden. Overall dementia severity assessments are designed to assign a level of severity to a patient’s condition, and are especially useful in assorting cases in research or service development. This paper considers scales used for each of these areas.

Cognition: screening for dementia

Scales in this section are included as they are used in clinical or research settings to screen for dementia, are brief (under 30 min), involve professionals interacting with patients and have been either recommended in reviews or guidelines [Brodaty et al. 2006; Holsinger et al. 2007; Milne et al. 2008; Appels and Scherder, 2010], or widely reported. Psychometric properties for each scale are summarized in Table 1. It should be noted that single cutoffs are never clearly best on any screening scale – those quoted have good combinations of sensitivity and specificity.

Table 1.

Short dementia screening tests suitable for primary and secondary care.

Instrument	Time to use (min)	Gold standard	Cutoff	Sensitivity	Specificity	Reference
MMSE	5–10	DSM-IV diagnosis	23/24	0.79	0.95	Hancock and Larner [2011]
AMTS	3–4	Clinical diagnosis	6/7	0.81	0.84	Antonelli Incalze et al. [2003]
Clock- drawing test	3	DSM III-R dementia	Shulman method, score 2/3	0.86	0.96	Brodaty and Moore [1997]
6-CIT	3–4	Clinical diagnosis of dementia	7/8	0.90	1.00	Brooke and Bullock [1999]
GPCOG	6	DSM-IV dementia	10/11 on total score	0.82	0.83	Brodaty et al. [2002]
Mini-Cog	3	Independent clinical diagnosis of dementia	Probably normal/ possibly impaired	0.76	0.89	Borson et al. [2003]
TYM	5–10	DSM-IV dementia	30/31	0.73	0.88	Hancock and Larner [2011]
MoCA	10	Clinical diagnosis of Alzheimer’s disease	25/26	1.00	0.87	Nasreddine et al. [2005]
ACE-R	15–20	DSM-IV dementia	73/74	0.90	0.93	Hancock and Larner [2011]
MIS	Under 5	Clinical diagnosis of dementia	5/6	0.86	0.91	Buschke et al. [1999]

Bold text indicates scoring direction of positive screen for dementia.

ACE-R, Addenbrookes Cognitive Assessment – Revised; AMTS, Abbreviated Mental Test Score; DSM, Diagnostic and Statistical Manual of Mental Disorder; GPCOG, General Practitioner assessment of Cognition; MIS, Memory Impairment Screen; MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment; TYM, Test Your Memory.

Abbreviated Mental Test Score

The Abbreviated Mental Test Score (AMTS) [Qureshi and Hodkinson, 1974] is a 10-item scale derived from a longer scale introduced previously [Hodkinson, 1972]. Any clinician can use this, and it takes only 3–4 min. It assesses orientation, registration, recall and concentration, and scores of 6 or below (from maximum of 10) have been shown to screen effectively for dementia, though as with many brief screens, low positive predictive values mean a second-stage assessment is always necessary [Antonelli Incalze et al. 2003]. Its brevity and ease of use have made it popular as a screening test in primary and secondary care nonspecialist settings.

Clock drawing

Numerous versions of the clock-drawing test have been devised, with many scoring algorithms [Brodaty and Moore, 1997]. Patients are typically asked to draw a clock face with numbers and hands (indicating a dictated time). It was designed as a quick and acceptable screening test for dementia. It is fast, requires no training and most scoring methods are fairly simple. It shows fairly good sensitivity and specificity as a screening test. It assesses only a very narrow part of cognitive dysfunction seen in dementia, and many other conditions (e.g. stroke) will affect it directly.

Mini-Cog

The Mini-Cog [Borson et al. 2000] is a very short test (3 min) suitable for primary care screening for dementia. It incorporates the clock-drawing test, adding a three-item delayed word recall task. It showed comparable sensitivity and specificity to the Mini-Mental State Examination (MMSE) in classifying community cases of dementia [Borson et al. 2003].

6-CIT

The 6-CIT [Brooke and Bullock, 1999] was designed for screening in a primary care setting. It takes 3–4 min to administer, and scoring is between 0 and 28, with cutoffs of 7/8 showing good screening sensitivity and specificity. It is easy to administer, though scoring is less intuitive than AMTS.

Test Your Memory

The Test Your Memory [Brown et al. 2009] test is a recently developed 10-item cognitive test designed to be self-administered under medical supervision. The maximum score is 50; at a score of 30 or below, the test has good specificity and sensitivity [comparable to MMSE and Addenbrookes Cognitive Assessment – Revised (ACE-R)] in distinguishing dementia from nondementia cases [Hancock and Larner, 2011]. This form of test may be attractive for time-limited clinicians wanting to screen for dementia, especially in primary care.

General Practitioner assessment of Cognition

The General Practitioner assessment of Cognition (GPCOG) [Brodaty et al. 2002] was designed for use in primary care and includes nine direct patient cognitive items, and six informant questions assessing change over several years. In total, it takes about 6 min. It has strong performance on sensitivity and specificity versus MMSE in detecting dementia in a typical primary care population [Ismail et al. 2009].

Memory Impairment Screen

The Memory Impairment Screen is a very brief four-item scale taking under 5 min to administer, and showing good sensitivity and specificity in classifying dementia [Buschke et al. 1999]. It lacks executive function or visuospatial items. Its use is likely to be confined to primary care, as an alternative to GPCOG, 6-CIT, clock-drawing, Mini-Cog or AMTS.

Mini-Mental State Examination

The MMSE [Folstein et al. 1975] is by some way the best known and most widely used measure of cognition in clinical practice worldwide. This scale can be easily administered by clinicians or researchers with minimal training, takes around 10 min and assesses cognitive function in the areas of orientation, memory, attention and calculation, language and visual construction. Patients score between 0 and 30 points, and cutoffs of 23/24 have typically been used to show significant cognitive impairment. It is widely translated and used. A standardized version [Molloy et al. 1991] improves its reliability, and is probably most important for research settings. The MMSE is unfortunately sometimes misunderstood as a diagnostic test, when it is in fact a screening test with relatively modest sensitivity. It has floor and ceiling effects and limited sensitivity to change. This in theory should limit its wider use in detecting change in clinical work and in research studies, though in these contexts it is still widely used, and even advocated [NICE, 2006].

Montreal Cognitive Assessment

The Montreal Cognitive Assessment [Nasreddine et al. 2005] was originally developed to help screen for mild cognitive impairment (MCI). It takes minimal training and can be used in about 10 min by any clinician. It assesses attention/concentration, executive functions, conceptual thinking, memory, language, calculation and orientation. A score of 25 or lower (from maximum of 30) is considered significant cognitive impairment. It performs at least as well as MMSE, including in screening for dementia. It has been widely translated. As it assesses executive function, it is particularly useful for patients with vascular impairment, including vascular dementia.

Addenbrookes Cognitive Assessment

The ACE [Mathuranth et al. 2000] and its commonly used revision the ACE-R [Mioshi et al. 2006] was originally developed as a screening test for dementia which, unlike the MMSE, would rely less on verbal than on executive abilities. It takes 15–20 min to administer and includes the items which lead to a MMSE score. It has been shown to have very high reliability and excellent diagnostic accuracy, and it is a practical option for clinical services intent on precision in diagnoses.

Longer cognitive assessments

Alzheimer’s Disease Assessment Scale – Cognitive section

The Alzheimer’s Disease Assessment Scale – Cognitive section (ADAS-Cog) [Rosen et al. 1984] is a detailed cognitive assessment for dementia, and takes a trained interviewer about 40 min to administer. It covers all cognitive areas in dementia and has good sensitivity to change.

The length of the assessment makes it generally unsuitable for clinical settings, but it is included as it is the leading assessment of cognitive change in drug trials in dementia, with a four-point difference between treatment groups considered clinically important [Rockwood et al. 2007].

Cambridge Assessment of Memory and Cognition

The Cambridge Assessment of Memory and Cognition [Roth et al. 1986] is the cognitive section of the comprehensive CAMDEX assessment. It covers a range of cognitive functions, including orientation, language, memory, attention, praxis, calculation, abstract thinking and perception. It takes around 25–40 min for a clinician to administer and requires a modest degree of training. It performs well against MMSE with no ceiling effects and conventional cutoffs of 79/80 have demonstrated excellent sensitivity and specificity for dementia [Huppert et al. 1995]. Its combination of breadth and relative brevity make it suitable for clinical use, particularly new assessments of patients in memory clinics. It has the added advantage of including questions to generate an MMSE score.

Function

Bristol Activities of Daily Living Scale

The Bristol Activities of Daily Living Scale (BADLS) [Bucks et al. 1996] was designed specifically for use in patients with dementia and covers 20 daily living activities. It takes a carer (professional or family) 15 min to administer. It is sensitive to change in dementia and short enough to use in clinical practice (carers may fill it in while clinicians are performing direct assessment of patients). It is regularly used as an outcome measure in clinical trials, where it is world leading as a dementia-specific measure. This outcome is among those recommended by a consensus recommendation of outcome scales for nondrug interventional studies in dementia [Moniz-Cook et al. 2008].

Barthel Index

The Barthel index [Mahoney and Barthel, 1965] is probably the best known assessment of functional ability for older people. It takes 5 min of informant’s time and has been widely translated and validated. It focuses on physical disability in 10 domains and should not be used other than to assess physical functional deficits in people with dementia, among whom cognitive deficits tend to confound assessment.

The Functional Independence Measure

The Functional Independence Measure [Keith et al. 1987] measures overall disability. It is observer rated and covers multiple important domains, including self-care, sphincters, mobility, communication, psychosocial function and cognition. Some training is required for its use. A UK version is available and it has been used in repeated observations of inpatients in general hospital [Zekry et al. 2008]. It is therefore an example of a scale which addresses cognitive as well as physical function, and is likely to be especially useful in inpatient or rehabilitation settings.

Instrumental Activities of Daily Living

The Instrumental Activities of Daily Living scale [Lawton and Brody, 1969] takes 5 min for a basically trained interviewer to assess ability in eight complex daily living tasks such as telephone use, shopping, housekeeping and finances. These abilities are more complex than the more basic abilities assessed by the Barthel scale, and therefore more sensitive to the cognitive changes seen in dementia. It is very commonly used in European memory clinics [Ramirez-Diaz et al. 2005].

The Informant Questionnaire on Cognitive Decline in the Elderly

The Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE) [Jorm and Jacomb, 1989] is a questionnaire administered to an informant outlining changes in everyday cognitive function. It aims to establish cognitive decline independent of premorbid ability by concentrating on 16–26 (depending on version) functional tasks, including recall of dates/conversations/whereabouts of objects, handling finances and using gadgets. It takes about 10 min to administer, and is conventionally used at the assessment stage in diagnosing dementia, usually combined with a direct cognitive assessment of the patient. This combination increases accuracy of diagnoses versus cognitive assessment alone [Jorm, 1994]. It is therefore suitable as a screening tool rather than in assessing change in function.

Behaviour

Neuropsychiatric Inventory

The Neuropsychiatric Inventory [Cummings et al. 1994] assesses a wide range of behaviours seen in dementia for both frequency and severity. These include delusions, agitation, depression, irritability and apathy. The scale takes 10 min for a clinician to administer to a carer. It has good psychometric properties and is widely used in drug trials, while being short enough (especially with patients without a wide range of behavioural issues) to consider for use in clinical practice.

Cohen-Mansfield Agitation Inventory

The Cohen-Mansfield Agitation Inventory [Cohen-Mansfield, 1986] takes 15 min for carers to rate, but requires some training. Up to 29 behaviours seen in dementia are rated for frequency – the lack of focus on severity is corrected by the breadth of behaviours covered. The behaviours covered include many of those found most disruptive, including verbal aggression, repetitiveness, screaming, hitting, grabbing and sexual advances. It is most commonly used in research settings.

BEHAVE-AD

The BEHAVE-AD [Reisberg et al. 1987] takes 20 min for a clinician to use, and is therefore most commonly used in interventional research studies. It covers most of the important disruptive behaviours, including aggression, overactivity, psychotic symptoms, mood disturbances, anxiety and day/night disturbances. Respondents are asked about the presence of behaviours and how troubling they are. It is reliable, sensitive to change and to stage of disease.

Quality of life

Generic measures of quality of life

EuroQol

The EuroQol measure [EuroQol Group, 1990] is a short, freely available generic measure of health-related quality of life. It can be simply administered to patients or carers in the form of a very brief self-completed questionnaire. There are two core components to the instrument: a description of the respondent’s own health using a health state classification system with five dimensions, and a rating on a visual analogue thermometer scale. It takes 2 min to complete. Like many quality of life instruments, carer and proxy ratings diverge widely, many patients with dementia cannot fill out the instrument, and the chief use of EuroQoL in dementia is as a health utility measure for measuring the economic impact of interventions in trials.

Short Form-36

The Short Form-36 (SF-36) [Ware and Sherbourne, 1992] and its shorter descendant the SF-12 [Ware et al. 1996] are examples of generic measures of quality of life which use recall over particular periods of time (typically 1 or 4 weeks) and are used to estimate health burden in large populations. These instruments have been shown to have high rates of noncompletion among frail older people and especially among those with moderate to severe dementia. They may have limited use for carers of people with dementia, but probably cannot routinely be used in practice with patients.

Dementia-specific quality of life instruments

Alzheimer’s Disease-related Quality of Life scale

The Alzheimer’s Disease-related Quality of Life scale (QoL-AD) [Logsdon et al. 1999] is a 13-item scale which has been extensively validated, is disease specific, can be completed by patient or carer and is suitable for use across the range of severity of dementia [Hoe et al. 2005]. It takes 10–15 min to administer. Patient and proxy versions are available. In a controversial area, its disease- specific properties, along with those of the health-related quality of life in dementia instrument (DEMQOL), make it a leading choice if quality of life is to be assessed [Moniz-Cook et al. 2008].

DEMQOL

DEMQOL [Smith et al. 2007] is a 31-item, disease-specific instrument for evaluating health-related quality of life in dementia, which shows comparable psychometric properties to the best available instruments and has been validated in a UK population. It has both patient-completed and proxy forms. Like QoL-AD, it is primarily likely to be used in research studies.

Depression in dementia

The Geriatric Depression Scale

The Geriatric Depression Scale (GDS) [Yesavage et al. 1983] is the most commonly used assessment of depressed mood among older people, and has been shortened to numerous versions, including a popular 15-item version (GDS-15) [Sheikh and Yesavage, 1986]. GDS-15 is usually self rated though can be rated by an assessor. It is sensitive to change and is reliable in older people in institutional care. It takes about 5–10 min to administer. Its major drawback in dementia is that it has been validated for people with mild dementia, but not for those with moderate to severe dementia (among whom completion rates may be low due to difficulty comprehending questions).

Cornell Scale for depression in dementia

The Cornell Scale [Alexopoulos et al. 1988a] is a 19-item scale in which questions are asked of the patient and the carer, meaning that the patient does not need to be able to answer questions for it to be used. The maximum score is 38. It has been validated patients with and without dementia [Alexopoulos et al. 1988b]. In patients with dementia, it is considered the gold standard for quantifying depressive symptoms.

The Montgomery Asberg Depression Rating Scale

The Montgomery Asberg Depression Rating Scale (MADRS) [Montgomery and Asberg, 1979] takes about 15–20 min for a trained assessor to complete. It is useful among older patients in that mainly psychological rather than confounding physical symptoms are assessed. It is particularly sensitive to change and often used in interventional research but the same issues as with GDS will limit its usefulness outside mild dementia.

The Hamilton Depression Rating Scale

The Hamilton Depression Rating Scale [Hamilton, 1960] is one of the most commonly used depression rating scales. It requires 20–30 min of questions in a semi-structured interview by a trained interviewer, and is therefore unlikely to be used in people with dementia. It is commonly used in antidepressant drug trials, and like MADRS, has a preponderance of psychological rather than physical items.

The Hospital Anxiety and Depression Scale

The Hospital Anxiety and Depression Scale [Zigmond and Snaith, 1983] is a popular screening test for depression and anxiety which was originally aimed at patients in hospital, though it has been used much more widely in recent years. It takes 3–5 min and is self-reported. Though easy to use and accurate at detecting depression, it has little practical use for older patients with significant cognitive impairment.

Carer burden

General Health Questionnaire

The General Health Questionnaire, 12-item version [Goldberg and Williams 1988] is a short self-rated scale designed to screen for psychological distress in the community. It is probably the most widely used and validated self-rated instrument for detection of psychological morbidity. It takes only a few minutes to administer.

Zarit Burden Interview

The Zarit Burden Interview [Zarit et al. 1980] is a 22-item self-report inventory of direct stress to carers in caring; it was designed for carers of people with dementia and has demonstrated sensitivity to change. Being disease specific gives it primacy in the area.

Overall dementia severity

Clinical Dementia Rating

The Clinical Dementia Rating scale [Morriss, 1993] allows more reliable staging of dementia than MMSE, and is based on caregiver accounts of problems in daily functional and cognitive tasks. It takes only a few minutes for clinicians already familiar with individual cases, and classifies people with dementia into questionable, mild, moderate and severe.

Global Deterioration Scale

The Global Deterioration Scale [Reisberg et al. 1982] is essentially for staging dementia and takes only 2 min once relevant clinical information has been collated. It has been well validated and classifies cases into seven stages from no complaints through to very severe. Like CDR, it is mainly used to assort cases by severity in research or in service development, as in an individual case, more subtle changes which are important may not be picked up.

Clinicians Global Impression of Change

The Clinicians Global Impression of Change (CIBIC-Plus) [Schneider et al. 1997] is a comprehensive global measure of detectable change in cognition, function and behaviour, usually requiring separate interviews with patients and carers. It is therefore conceptually attractive for assessing progression, but requires a trained clinician and 10–40 min of interview time so may be unsuited to routine clinical practice.

Discussion

A key consideration in deciding what dementia assessment scales to choose is to clarify the question being asked. Consensus guidelines have been attempted [Ramirez Diaz et al. 2005; Moniz-Cook et al. 2008]. Most of the brief screening instruments like 6-CIT, clock drawing and AMTS are probably psychometrically as good as a common instrument like MMSE in screening for significant cognitive impairment, and are a little shorter. They lack the breadth of assessments in MMSE and are therefore to be used only in settings in which time or frailty make longer assessment impossible. The diagnosis of dementia is always based on a clear history and invariably involves collateral history from an informant along with direct patient assessment. Some comprehensive instruments to aid this diagnosis have been developed. In memory clinics, structured neuropsychological assessment and the use of IQCODE to detail cognitive change as observed by a carer are often used to improve precision of diagnostic decisions. In borderline or mild cases of dementia, assessments probably need to include assessments of at least this complexity, with important guidelines explicitly recommending this [NICE, 2006]. Such assessments will usually involve assessment of premorbid ability and quantification of explicit cognitive deficits, including, but not limited to, memory, and establishing deficits compared with expected norms. Commonly, these specialist assessments involve a specially trained neuropsychologist. Scales like the ACE-R can easily be used in clinical settings by clinicians other than neuropsychologists. In monitoring progress over time, cognition (for example with MMSE, though subject to ceiling/floor effects and relatively insensitive to change), function (e.g. BADLS) or a generic measure of overall severity of dementia(e.g. Clinical Dementia Rating, Global Deterioration Scale, CIBIC) are often used. If cognitive performance is of specific interest, a well validated scale like ADAS-Cog is preferred, despite its length. For clinical trials in which cognition is of primary interest, a de facto gold standard of a four-point change on ADAS-Cog has been established [Rockwood et al. 2007]. In assessing depression in dementia (Cornell scale) and carer stress (Zarit Burden Inventory) there are relatively clear leading assessment scales. Quality of life assessment in dementia is a minefield due to the disparity between patient and proxy ratings, and poor completion rates with more severe dementia. The recent introduction of dementia-specific scales for quality of life, which allow proxy ratings, is at least a significant step forward. Assessing change in behavioural symptoms in dementia is especially important in judging treatment effects (for example – has the patient improved during short-term treatment with antipsychotic medication enough to justify risks of continued prescribing?). Well established scales are available for this purpose. A general principle in dementia assessment is that disease-specific instruments are usually markedly superior in clarifying judgements made. Subject to limitations on clinical and research resources, these instruments should be considered first to maximize clinical practice. A great deal of effort goes into choosing and justifying primary outcome measures in research trials; as clarity about intervention effects is so important, basic familiarity with the strengths and weaknesses of commonly used assessment scales in dementia can help improve the rigour of clinical practice.