Diagnosis of dementia

BMJ 2009; 338 doi: http://dx.doi.org/10.1136/bmj.b1176 (Published 10 June 2009) Cite this as: BMJ 2009;338:b1176
  1. Claire Nicholl, consultant physician
  1. 1Medicine for the Elderly, Addenbrooke’s Hospital, Cambridge CB2 2QQ
  1. claire.nicholl{at}addenbrookes.nhs.uk

    The usefulness of screening tests varies according to the clinical setting

    Although dementia is common, with an expected prevalence of 13 in 1000 in people aged 65-69 and 122 in 1000 in those over 80, only about half of those affected are diagnosed.1 Without a diagnosis, patients and carers cannot access the services they need, so earlier diagnosis is a key component of the National Dementia Strategy in the United Kingdom.2 3 However, early diagnosis is not easy and no definitive test exists. In the linked cross sectional study (doi:10.1136/bmj.b2030), Brown and colleagues assess the effectiveness of the “test your memory” cognitive test in detecting Alzheimer’s disease. The test was designed to minimise operator time and to be suitable for non-specialists to use.4

    Cognitive function in established dementia is clearly different from the cognitive changes of normal ageing, but the onset of dementia is gradual. Diagnosis requires consideration of the person’s education, culture, and circumstances; dementia presents when a person’s cognitive abilities are no longer adequate for them to cope with their environment. A high flying executive may experience problems early in the neuropathological process, whereas a resident of a care home may have no difficulty with his or her routine until pronounced changes have occurred. Factors including English as a second language or coexisting depression make the diagnosis even more complex.

    Should we screen or case find? This is not explicit in the National Dementia Strategy, although it does seem to promote screening.5 Case finding requires a high index of suspicion and assessment if the affected person or an informant expresses concern about any aspect of cognition, which usually but not always involves memory.6 Other pointers may be a deterioration in personal appearance, a reduction in social roles, or when things simply “don’t fit.” The general practitioner may note unexplained erratic international normalised ratios in a patient whose anticoagulation was previously not problematic. In secondary care, delirium after acute illness or surgery warrants follow-up.

    The diagnosis of dementia requires a detailed history from the patient and an informant, in addition to mental state examination and cognitive testing, usually in a specialist setting. However, a quick screening test is needed in primary care and general hospital practice. The ideal test would be sensitive and specific (parameters that vary with the prevalence of dementia in the population being tested), have face validity, and have inter-rater and test-retest reliability.7 It would be independent of age, sex, education, mood, culture, and language; be quick and user friendly; and have an unambiguous scoring system. The pattern of errors would discriminate between dementia types and the proforma would be readily available and free. The test should also be sensitive to change and not have a ceiling or floor. Unsurprisingly, no test fulfils all these criteria.

    Different screening tests are suitable for different settings, and the plethora of tests available suggests that no one test is the best.8 There is generally a trade off between brevity and performance. Shorter tests may confirm a cognitive problem that needs to be evaluated, whereas longer tests contribute more to the diagnosis. The “worried well” can be reassured, but people with “mild cognitive impairment” need a management plan.9

    The appeal of Brown and colleagues’ test is that it is designed to be self administered, so a range of cognitive domains can be assessed with little demand on the clinician’s time.4 However, five of the 50 points relate to the amount of help needed to complete the test, so some assessor input is needed. In the diagnosis of early Alzheimer’s disease, with a cut-off point of ≤43/50, it had good sensitivity (92%), specificity (84%), and inter-rater reliability, and it was more sensitive than the mini-mental state examination (92% v 54%) in their population. The authors do not comment on the ethnicity of the study participants, but the local population is mainly white. Some of the items probably show cultural bias (labelling a man’s suit would not be a fair test for a Bangladeshi woman) and dating the first world war has been criticised in the abbreviated mental test score, but the authors promise a range of modified tests. People with visual impairment may have difficulty doing the test.

    Where will this test be useful in clinical practice? Recommended tests in primary care include the general practitioner assessment of cognition, the memory impairment screen, and the mini-cognitive assessment instrument.10 General practitioners are likely to continue to use these briefer scores. The abbreviated mental test is still used despite its limitations. The British Geriatrics Society recommends the mini-mental state examination with CLOX1, an executive clock drawing test where the subject draws a clock in response to verbal commands, and the informant questionnaire on cognitive decline in the elderly.11 However, this last test is affected by depression in the informant and the quality of the relationship between the patient and the informant. Clock drawing tests, such as CLOX1, screen for several cognitive functions rapidly and unobtrusively, but an important detractor is the multiplicity of scoring methods.

    The most commonly performed test is the mini-mental state examination, which is used in research and clinical settings; test results are used as a guide to starting and monitoring treatment with anticholinesterase inhibitors in England.12 Longer scales are used in specialist settings, but the test your memory test is not designed to replace these.8 9

    If the test your memory test is to be adopted more widely it must be validated in a range of settings and different populations. Until then, the most important message is that clinicians should identify a test that suits their clinical setting, use it to screen or case find as appropriate, and develop experience in its use to improve the identification of patients with early dementia.


    Cite this as: BMJ 2009;338:b1176