Arshad et al (Reference Arshad, Sridharan and Brown2001) raise important concerns that UK guidelines for the treatment of Alzheimer's disease (National Institute for Clinical Excellence (NICE), 2001) may be counterproductive for patients with learning disabilities. Potential for discrimination does not by any means stop here. A particular difficulty they highlight is the central role of the Mini-Mental State Examination (MMSE) instrument in determining treatment ‘eligibility’ and response. Scores on the MMSE are strongly influenced by previous education and cross-cultural validity is poor. The guidelines are, therefore, unhelpful for people with lower educational attainment or for growing numbers of older people from minority ethnic groups in the UK. Comorbid cerebrovascular disease will also be more frequent in people from more disadvantaged backgrounds and, in particular, minority ethnic groups such as African—Caribbean populations (Reference Stewart, Dundas and HowardStewart et al, 1999). This reduces the likelihood of a diagnosis of Alzheimer's disease (and therefore eligibility for anti-cholinesterase treatment) according to standard diagnostic criteria (Reference McKhann, Drachman and FolsteinMcKhann et al, 1984), despite growing evidence for overlapping pathological processes in dementia (Reference Holmes, Cairns and LantosHolmes et al, 1999).

For sub-populations who are under-represented in clinical trial samples (minority ethnic groups, people with lower educational attainment, people with learning disability, people with comorbid cerebrovascular disease), the best that can be hoped for is that a considerably weaker evidence base might emerge some years in the future. By this time large numbers of people may have failed to receive potentially beneficial treatment. The problem does not lie with treatment guidelines themselves but with how they are applied at the level of individuals and services — in particular regarding groups with Alzheimer's disease for whom a 26-year-old cognitive screen and/or 17-year-old diagnostic criteria are unhelpful. Evidence-based medicine is a noble ideal. However, clinical practice that is restricted to the evidence base may amount to institutionalised discrimination.