There are limitations with this classification. Only 10-15% of patients diagnosed with MCI convert to dementia per year in clinical populations and 5-10% in the community.

A substantial proportion are diagnosed as normal at follow up.  The criteria are circular to some extent (ie. the absence of dementia) and there has been criticism of the focus on memory impairment as the exclusive cognitive deficit because some dementias begin with other cognitive deficits or personality change.  For example, fronto-temporal dementia begins with changes in behaviour (eg. disinhibition, apathy, obsessive compulsive behaviour) and semantic dementia begins with language problems. Nevertheless, the MCI diagnosis fills a gap that is important both clinically and for researchers, and until an objective test of preclinical dementia is developed, it remains a necessary classification.

Differentiating normal changes from dementia
Unfortunately, there are no hard and fast rules to easily distinguish between normal aging and dementia.

Differentiating normal cognitive aging from MCI and early stage dementia requires formal assessment by a neurologist,  geriatrician or psychogeriatrician,  and ideally, a neuropsychologist. Screening instruments are available for use in general practice,  but these only yield a cut-off score that is associated with an established risk in population based studies. The MMSE is a widely used instrument in which scores under 24 are generally taken to indicate cognitive impairment.

However, the scale is inaccurate, with highly educated patients less likely to be diagnosed and patients from non-English speaking backgrounds more likely to be diagnosed. This is because parts of the test are based on English literacy and numeracy skills. Another instrument available in Australia is the GPCOG,  which has been developed for use by GPs. This combines cognitive testing with informant interview and takes 4-6 minutes to administer.  When validated in an Australian primary care population,  the GPCOG performed with a sensitivity of 0.85,  specificity of 0.86, misclassification rate of 14%,  and positive predictive value of 71.4%.  It appears to be free of an education bias, but has not been tested for language and culture biases.

Key considerations for diagnosis of a cognitive disorder include:

•   a history of decline in cognitive abilities
•   evidence that this decline interferes with everyday activities, and
•   impairment in more than one cognitive domain.

Interviewing an informant is the best way of obtaining an accurate history.  Patients who present a diagnostic challenge include those with low education, premorbid intellectual disability, those currently suffering another mental health condition such as depression or psychosis, and those from a non-English speaking background.