Finally, some good news!
Professor Carol Brayne of Cambridge University has rocked the field with a result that contradicts one of our basic assumptions – that dementia rates are on the rise. Using data from the 20-year Cognitive Function and Ageing Study (CFAS), her team has reported in The Lancet, that age-specific dementia prevalence rates have fallen by 20-30% in the period between 1991 to 2011. In this special article, RNG leader Michael Valenzuela interviews long term collaborator Prof Brayne about this transformational work.
[MV] Would it be correct to summarise your main result to be that your average 80-year old Brit born in 1931 has about a 30% lower chance of having dementia today (6.5%) than your average 80-year old Brit in 1991 who was born in 1911 (8.3%)?
[Carol]: The gist is correct although the exact amount might not be quite right and we would usually frame it with the uncertainty (the range of reduction possible), but the message is certainly that the risk of having dementia has gone down for most ages in both sexes.
[MV] So at last some good news on the dementia front?
[Carol]: Given the findings from our and other studies, both in Europe and US these do seem to be encouraging findings.
[MV] When speculating about the possible drivers of this generational change, you suggest secular changes in education and better management of vascular risk factors as possible explanations. Did CFAS I and II differ on any of these risk factors?
[Carol]: We have been concentrating on the analysis of the prevalence studies and will move on to the analysis of the factors (insofar as we can and have measured them) which might be influencing our findings. There is much more work to be done!
[MV] Seems to me like the largest secular changes have perhaps occurred more recently – after these individuals retired – than when they were in high school age (e.g., 1929 vs 1949). What’s your view on this?
[Carol]: I’m not sure whether this is true and of course it depends on how one defines large! There have been major changes on a range of fronts from early life education, nutritional patterns throughout the lifecourse, the nature of midlife exposures (the waning of the smoking epidemic in men for example) and large changes in mid to late life vascular risk and outcomes, length of life itself and then the changes to expectations of lifestyle in retirement…So I think it’s difficult to pre-judge the relative impact of any such changes.
[MV] Your Figure 2 shows there is a lot more geographical variability in dementia rate in older women than men. In fact, the pattern looks similar between sexes, but just ‘shifted to the right’ in women. Maybe there is more room to move in the dementia rate in women?
[Carol]: This is also something we will be exploring in more detail as the basic differences here are affected more by the finding that there is variation according to deprivation (or poverty) indicators.
[MV] Do you think an appropriate ‘dementia prevention’ target could be to strive to equalise rates between the sexes, and strive for all regions to target the lowest rate nationally?
[Carol]: There are known variations in the survival of women in the presence of greater disability generally than men so this might be rather difficult to achieve. The underlying reasons for this survival differential are not known and I don’t think a target would necessarily be helpful in this regard. What would be good would be to have societal structures and policies that optimise health and wellbeing over the earlier life course so that folk reach 65 without major non communicable disease, which would at least mean they enter older ages with lower risk.
[MV] Having worked with the CFAS data myself, I still don’t think I fully understand the mysteries of the ‘weightings’ used to adjust the prevalence estimates! How confident are you that the result is not an artefact of these weightings, or more importantly, changes in primary care registration patterns on which the sampling is based?
[Carol]: This is a reasonable question. We were very aware of the societal differences once we embarked on the attempt to replicate the study 20 years on. We have done as much as we can to make sure that our results are not driven by these changes. It would be wonderful if the next time we attempt to look at time trends that we have more accurate population enumeration and higher response rates. However, the trends are not in this direction at present.
[MV] Can we also clarify the distinction between estimated dementia prevalence rates (I.e., % of a given population) and estimated number of individuals with dementia? In your paper, based on CFASII prevalence rates and current UK age and sex structure, you calculate a current dementia burden of 670,000. You then cite a number of 664,000 based on CFAS I (1991) data. But your previous published estimate was 543,000 (CFAS Psych Med 28 (1998); 319-335). Why the change in CFAS 1991 burden numbers?
[Carol]: It is important to be clear about which population is being estimated for – England, UK etc. For the comparisons now of the earlier study (MRC CFAS) with the new one we created a ‘new’ old study from three of the original sites. These were the three we had selected out (to provide geographical variation as well as including the variation that we’d seen first time round, even though not significant in the earlier analyses). So we ended up with CFAS I and CFAS II which provided a direct comparison of the findings from the first study in those three sites and the second. These were analysed in the same way which is why the results are slightly different.
[MV] Related to above, my guess is that while a cohort change may be occurring, there are whole lot more people entering the older age bands, and so dementia numbers are likely to be on the rise. What I think your data teaches us is that the pace of any increasing burden is probably not as stark as we were assuming. Is that correct?
[Carol]: The fact that the estimates of dementia for the UK are effectively stable despite the reduction at most ages in the prevalence of dementia is because of the change in the age structure of the population. We have previously shown in healthy life expectancy analyses that there is a relatively fixed proportion of life spent with cognitive impairment after the age of 65 and it is possible that the phenomenon we are seeing here is the strong relationship of dementia to death with the implication that we need to be cautious about the true extent of the preventability of dementia over the lifecourse (as opposed to at any given age). Further research should help tease these issues out.
[MV] Perhaps those with the greatest to lose are the NGOs and advocacy groups who have been relying on the tsunami argument. In fact, we all have to one degree. Are you off their Christmas list?
[Carol]: After a whole career in dementia research I hope not! I think our findings serve to emphasize the importance of taking the big picture into account. Dementia remains a major concern for ageing societies and the message is one of balance of investment in research and action across the lifecourse.
[MV] Thanks so much and congratulations to you and your team on this fundamental work.
[Carol]: Thanks!