As part of a series of lectures on medical statistics, Emmanuel College hosted a talk by Fiona Mathews about her research into estimating the extent of dementia in the UK population. I have an interest in recently developments in medical research and encouraged by a friend decided to go along to find out what it was about.
In our ageing society, dementia is becoming a serious problem – with over half of people dying over the age of 85 having dementia. However, very little progress has been made on medical treatments or even in understanding the pathology of this disease. While society and government’s apparent lack of concern for mental and elderly health could be a major contributor towards this, another issue is simply that it is very challenging to get accurate statistics on the extent of dementia in the population and how is it changing over time. Dr Mathews’ talk focused on these challenges.
The first major challenge in estimating what proportion of the population have dementia is one of diagnosis – quite simply there is no definitive test to conclusively confirm whether or not someone has the disease. Instead, individuals are diagnosed based on discussions with their family over behavioural changes, physical and cognitive exams, brain imaging scans and blood tests to identify potential biomarkers. However, since dementia is defined as a change in the brain but we rarely have scans available of each patient’s brain at a younger age to compare against, in many cases experts disagree on a diagnosis. When it comes to analysing statistics therefore, how do we know if we can trust the data? One method used is to use a fixed diagnostic test (e.g. looking for particular biomarkers), even if it is not 100% accurate, and observing how this non-subjective measure of dementia varies in the population over time.
However, even if we have a way of diagnosing dementia this then brings up the next challenge – that of participation. The Cognitive Function and Ageing Study (CFAS) noticed a decrease in participation rates from 80% down to barely over 50% between their two phases in 1994 and 2011. Therefore, although the study indicated an overall decrease in the prevalence of dementia, the reduced sample size decreased the reliability of the data and there are concerns that the reduced participation may have been biasing the results, for example if people with dementia in 2011 are less likely to take part than those in 1994 due to lifestyle changes in that time. More and more, elderly people are busy with events, grandkids, travelling and in some cases work – and if those with dementia are for some reason participating more in these things or are more sick due to a change in the severity of the disease, they are less likely to participate in the study than before.
A further challenge is knowing which statistical sources to look at. Death certificates only list causes of death, not all diseases suffered by the patient at that time, thus missing most dementia cases. Dementia cases should be known by GPs, but the current NHS target of diagnosing 67% of dementia cases (using a predicted number calculated from phase 1 of CFAS) may mean doctors are not recording all cases accurately. Again, the issue of subjectivity arises – if doctors are more educated about dementia than in the past they may be more likely to diagnose a patient with it, leading to the statistics indicating an increase in the incidence of dementia when in fact it was just better reporting of the disease.
The fourth challenge Dr Mathews’ discussed was that of understanding the causes of dementia. Most of the factors associated with dementia, such as hypertension and inactivity, are linked to cardiovascular mechanisms. However, these each are in turn linked to each other so it cannot simply be said what is the underlying cause. Taking these interactions between factors into account, only 30% of dementia cases are linked to the main risk factors for the disease. Therefore, statistics cannot give us a perfect picture of the underlying mechanisms of dementia.
The final challenge mentioned was deciding what metric to use – fatality or new cases? If the fatality of dementia is decreasing, for example due to improved healthcare, more people with the disease survive for longer, thus increasing the apparent prevalence of the disease despite the fact that the number of new cases may have remained the same. In order to get a more accurate picture, the number of new cases is a better measure – which can be done by only looking at the results for those individuals who participated in both the 1994 and 2011 surveys. This analysis indicated that the decreased prevalence of dementia was not due to an increased mortality and so appeared to be an actual reduction in new cases.
In conclusion, the percentage of the population suffering from dementia appears to be decreasing, although various challenges prevent an exact percentage from being given. However, even if it was assumed that half of those who didn’t participate in the 2011 study compared to the 1994 one had dementia the statistics still show an overall decrease in new cases. Despite this, it is still important that more research is done into the mechanisms of this disease and into finding new treatments. Overall, a very insightful talk which made me think more about the importance of statistics and data in directing research funding and public policy.