Dementia incidence trend in England and Wales, 2002-19, and projection for dementia burden to 2040: analysis of data from the English Longitudinal Study of Ageing

Abstract

Background: Dementia incidence declined in many high-income countries in the 2000s, but evidence on the post-2010 trend is scarce. We aimed to analyse the temporal trend in England and Wales between 2002 and 2019, considering bias and non-linearity.

Methods: Population-based panel data representing adults aged 50 years and older from the English Longitudinal Study of Ageing were linked to the mortality register across wave 1 (2002-03) to wave 9 (2018-19) (90 073 person observations). Standard criteria based on cognitive and functional impairment were used to ascertain incident dementia. Crude incidence rates were determined in seven overlapping initially dementia-free subcohorts each followed up for 4 years (ie, 2002-06, 2004-08, 2006-10, 2008-12, 2010-14, 2012-16, and 2014-18). We examined the temporal trend of dementia incidence according to age, sex, and educational attainment. We estimated the trend of dementia incidence adjusted by age and sex with Cox proportional hazards and multistate models. Restricted cubic splines allowed for potential non-linearity in the time trend. A Markov model was used to project future dementia burden considering the estimated incidence trend.

Findings: Incidence rate standardised by age and sex declined from 2002 to 2010 (from 10·7 to 8·6 per 1000 person-years), then increased from 2010 to 2019 (from 8·6 to 11·3 per 1000 person-years). Adjusting for age and sex, and accounting for missing dementia cases due to death, estimated dementia incidence declined by 28·8% from 2002 to 2008 (incidence rate ratio 0·71, 95% CI 0·58-0·88), and increased by 25·2% from 2008 to 2016 (1·25, 1·03-1·54). The group with lower educational attainment had a smaller decline in dementia incidence from 2002 to 2008 and a greater increase after 2008. If the upward incidence trend continued, there would be 1·7 million (1·62-1·75) dementia cases in England and Wales by 2040, 70% more than previously forecast.

Interpretation: Dementia incidence might no longer be declining in England and Wales. If the upward trend since 2008 continues, along with population ageing, the burden on health and social care will be large.

Funding: UK Economic and Social Research Council.

Figure 1

Age-adjusted and sex-adjusted dementia incidence rate ratio estimated from the (A) Cox model and (B) multistate model, 2002–19 Reference year 2002. Dashed lines represent 95% CIs. The multistate model (appendix p 7) accounted for potential bias due to deaths between waves among newly incident dementia cases. The incidence rate for the final wave 2018–19 could not be estimated. Cox model: hazard ratio 0·85, 95% CI 0·74–0·99 for year 2010 versus 2002, and 1·20, 1·04–1·39 for year 2014 versus 2010. Assuming a linear trend after 2010, dementia incidence increased by 4·6% annually. Multistate model: 0·71, 0·58–0·88 for year 2008 versus 2002, and 1·25, 1·03–1·54 for year 2016 versus 2008. Assuming a linear trend after 2008, dementia incidence increased by 2·8% annually.

Figure 2

Dementia incidence rate by age in (A) men and (B) women at years 2002, 2008, and 2016 Rates were estimated from the multistate model using English Longitudinal Study of Ageing data, 2002–19.

Figure 3

Age-adjusted and sex-adjusted dementia incidence ratio by (A) age, (B) sex, and (C) educational attainment Rates were estimated from the multistate model using English Longitudinal Study of Ageing data, 2002–19. Reference year 2002. p=0·78 for age, 0·84 for sex, and 0·0086 for educational attainment.

Figure 4

Projected number of people with dementia in England and Wales, 2018–40 Shaded areas represent 95% uncertainty intervals. 2·8% relative annual increase (black line) was estimated using English Longitudinal Study of Ageing data, 2002–19, based on a multistate model accounting for non-linear trend and bias. 2·7% relative annual reduction (red line) was based on a previous estimate using English Longitudinal Study of Ageing data, 2002–12.