Background The secular trend of dementia incidence is an important input parameter in projecting the future burden of dementia in a population. The estimated disease burden informs health and social care planning, and prevention policy. In the context of rapidly changing aging and health policy development in China, we hypothesize that the calendar effect of dementia incidence may not be linear. We used the Chinese Longitudinal Healthy Longevity Survey (CLHLS) to estimate the secular trend of dementia incidence in China, taking into account potential nonlinear trend.

Methods In this analysis, 6 waves (2002 to 2018) of the CLHLS cohort were used. Incident dementia was defined as onset of simultaneous cognitive impairment and functional impairment based on standard test items, or self-reported doctor diagnosis. Follow-up started at study entry and continued until incident dementia, loss to follow-up, death, or end of follow-up, whichever came first. We fitted three statistical models: Cox, Fine-Gray (additionally accounting for competing risk of death) and competing risk Joint model (additionally accounting for informative dropouts) to estimate the calendar effect of dementia incidence. All three models explored the possible nonlinearity of the calendar effect by using natural cubic spline functions.

Results Making a linear trend assumption, the estimate of age-sex-adjusted dementia incidence from the Cox model tended to increase by 1.4% annually (95% CI: -0.15% to 3.0%); the estimate from the Fine-Gray model was -0.4% (95% CI: -1.8% to 1.0%); the estimate from the Joint model was -0.3% (95% CI: -1.6% to 1.0%). Making a non-linear trend assumption, all three models showed a stable trend of dementia incidence from 2002 to 2010 and a declining trend after 2010.

Conclusion Our hypothesis of a non-linear trend of dementia incidence was confirmed in CLHLS. Robust statistical approaches, accounting for competing risk of death and informative dropouts, are needed to accurately estimate the secular trend of dementia incidence. We found that dementia incidence in China had a stable trend from 2002 to 2010 and had a declining trend after 2010.