Dementia trends in England and Wales.

Because the most prominent risk factor for dementia is aging, the urgent need to address the public health challenges of dementia is heightened by the “greying” of societies worldwide. In 2015, nearly 47 million people around the world had dementia. If age and sex specific prevalence of dementia stays constant over time, there will be 130 million cases by 2050. And yet changes over time in both dementia incidence (the rate at which new cases arise in a population over a specified period, among people at risk) and duration (time from the overt clinical onset to death) are plausible, so dementia prevalence, which is the product of incidence and average duration, could remain stable, decrease, or increase accordingly. In a linked paper, Ahmadi-Abhari and colleagues (doi:10.1136/bmj.j2856) develop a Markov model to estimate the direction and magnitude of these and other trends to 2040 in England and Wales. Trends in prevalence and incidence of dementia have been previously reported in the Medical Research Council’s Cognitive Function and Ageing Study (CFAS), which was purposely designed and powered to detect any such changes. The study found a reduction in incidence, although in men only, and a reduction in prevalence in those aged 80 or more years. However, the net effect of the incidence reduction is uncertain, and whether sex and age standardised prevalence of dementia is declining in Western countries is still unclear. Differential participation and attrition of participants cannot be ruled out because dementia risk at baseline is likely associated with non-participation and shorter survival. Dementia shares risk factors with other non-communicable diseases of old age, which might also increase all cause mortality. Modelling to account for missing data had no impact on reduction in dementia incidence over two decades in the CFAS I and II. However, uncertainty remains over whether and by how much changes in the prevalence of and mortality from vascular and other risk factors will modify the magnitude of the expected dementia epidemic. Using data from 1448 men and women from the English Longitudinal Study of Ageing (ELSA), Ahmadi-Abhari and colleagues found a statistically significant 2.7% annual reduction in dementia incidence between 2002 and 2013, after accounting for differential attrition. To estimate projections of dementia prevalence in England and Wales up to 2040, the authors applied advanced statistical methodologies that accounted for varying competing risks and mortality rates, and their potential opposite effect on changes in dementia prevalence over time. The novel and advanced statistical models and the extensive set of sensitivity analyses conducted to test their robustness are major strengths of this study. Although a 21% decline in the overall age standardised prevalence of dementia was estimated between 2016 and 2040, there will be up to a 57% increase in the absolute number of those affected by 2040 because of the expected demographic changes. These results are in line with those of previous studies. Some design characteristics and methodological limitations suggest that the results of this rich set of secondary analyses from the ELSA study should be interpreted cautiously. Dementia diagnosis requires demonstration of cognitive decline (from a previous level of functioning), demonstration that the decline socially and functionally impairs patients, and the exclusion of “other causes” (including depression and delirium). Although the criteria to establish dementia caseness were kept constant over time in the ELSA study, they were not formally validated, were not equal for all participants at each wave, and do not seem to conform to DSM-IV (diagnostic and statistical manual of mental disorders, fourth edition) or other criteria. Evidence of a major decline in cognitive function compared with the previous level of functioning was not sought in all participants; and dementia diagnosis based on self reported doctor diagnosis and the use of an informant questionnaire, which was used in the subsample of those with no cognitive assessment, were likely prone to bias and accounted for 32% and 17% of all dementia cases identified, respectively. Both cognitive and functional impairment may be due to depression, delirium, or other mental disorders, but these were not assessed. The possibility that observed declines in both incidence and prevalence of dementia might be in part (or entirely) due to a reduction in the occurrence of depression cannot be ruled out. Finally, although temporal trends in mortality due to dementia are difficult to study, and were not reported, it is possible that future longer survival of patients with dementia (which is desirable) may counterbalance the effects of a reduced incidence on prevalence.