Population attributable fractions for modifiable risk factors of dementia in cognitively normal and MCI populations: Data from two countries

Abstract Background Previous studies have shown that population‐wide up to 40% of all dementia cases are related to twelve modifiable risk factors, which thus calls for preventive measures. To inform health promotion strategies, it is important to know whether the risk attributable to modifiable risk factors differs in cognitively healthy persons and persons with Mild Cognitive impairment (MCI). This study analyzed associations and population attributable fractions (PAFs) of twelve modifiable risk factors with dementia in cognitively healthy and declining (MCI) adults aged ≥65 years, in two cohort studies from the US and Greece. Method Data were from the National Alzheimer’s Coordinating Centre (NACC) (n=16,147, impaired‐not‐MCI n=1007, mean age 75.2 ±6.9, 54.8% female) and the Hellenic Longitudinal Investigation of Aging and Diet (HELIAD) (n=1141, mean age 72.9 ±5.0, 58.0% female). PAFs were calculated based on hazard ratios (HRs) for the risk of dementia and risk factor prevalence in the NACC (9 risk factors) and HELIAD (10 risk factors) data. Weighted PAFs were calculated to adjust for communality between risk factors. Analyses were performed for the total samples, as well as separately for cognitively healthy participants and those with MCI at baseline. Result 2,630 participants enrolled by the NACC developed MCI (25.1%) and 3,333 developed dementia (20.7%) during a mean follow‐up of 4.9 (±3.5) years. 131 participants developed MCI (11.2%) and 68 developed dementia (5.9%) during HELIAD follow‐up of on average 3.1 (±0.86) years. A relatively low weighted overall PAF of 21.4% was found in the total sample of the NACC study. The PAF was 20.8% in the cognitively normal subgroup and ‐6.17% in the MCI subgroup A relatively high weighted overall PAF of 60.7% was found for the total sample in HELIAD. The PAF was 78.4% in the cognitively normal subgroup and 80.6% in the MCI subgroup. Conclusion In general, translating valuable global meta‐analysis data on modifiable risk factors should be carefully carried out per population before starting preventive interventions. Within each cohort, the PAFs for cognitively healthy and cognitively declining participants were in the same range, suggesting that also in persons with MCI, elimination of risk factors continues to be important.