John: Hello, I'm John Hilton, editor of the Cochrane Editorial unit. As the problem of dementia grows, so does the need to diagnose it early. Cochrane Reviews have examined various tests for doing so and a new Cochrane Review in November 2016 looks at one, called IQCODE. We asked lead author, Terry Quinn from the University of Glasgow in the UK to tell us more.

Terry: Dementia is one of the major challenges facing contemporary, global healthcare. An important first step in its management is making an early and accurate diagnosis and many assessment tools are available to help clinicians screen for dementia. Often, these involve using pencil and paper and/or memory tests with the patient, but we were interested in an assessment that takes a different approach. The Informant Questionnaire on Cognitive Decline in the Elderly (often abbreviated to IQCODE) focusses on someone that knows the patient, such as a family member or friend, and asks them standard questions to assess for change in memory over time.
In our other Cochrane Reviews, we’ve shown that IQCODE performs reasonably well as a dementia screening tool and we’ve now assessed whether the IQCODE could predict future cognitive decline in people who do not have frank dementia at the time of testing. At first, this may seem a counterintuitive question, but our hypothesis was that a low score on the IQCODE could signify the early stages of an underlying neurodegenerative condition that did not yet fulfil the formal diagnostic criteria for dementia.
We found three relevant studies, representing 626 patients. However, the people in these studies, the settings and the way the IQCODE was used were so diverse that we could not pool the data in a meta-analysis. All studies performed IQCODE at the point of a major physiological insult that could have a deleterious effect on cognition. Two concerned acute stroke and the other hip fracture. So, although all three studies were relevant to our study question, they do not tell us about the accuracy of IQCODE for the natural history of dementia that is not complicated by another illness.
Our findings need to be interpreted with caution, because the included studies all had potential for bias, particularly around loss to follow-up and baseline assessment for dementia. But, in general, we found that IQCODE was a sensitive predictive tool but lacked specificity. This means a 'positive' IQCODE picked up most patients who would go on to develop dementia, but mislabelled some who did not develop dementia, which might make it unsuitable for use on its own.
In summary, we found limited published evidence on the accuracy of IQCODE for predicting future dementia. Where data were available, IQCODE lacked specificity and may label a number of people as being ‘at risk of dementia’ when they are not. Because of this, we feel that IQCODE might not be a suitable predictive tool for dementia when used alone.

John: If you would like to find out more about the review’s findings and the conclusions of the authors, or to read the other reviews of IQCODE that Terry mentioned, just go online to Cochrane Library dot com and search 'iqcode’.