Population attributable fractions for risk factors for dementia in low-income and middle-income countries: an analysis using cross-sectional survey data.
Level 4 - case-series / case-control
Cross-sectional survey data combined with meta-analytic relative risks for population attributable fraction modeling
PubMed 31000129 · doi:10.1016/S2214-109X(19)30074-9
What was done
Researchers calculated population attributable fractions (PAFs) for nine potentially modifiable dementia risk factors (low childhood education, midlife hearing loss, hypertension, obesity, late-life smoking, depression, physical inactivity, social isolation, and diabetes) using cross-sectional survey data from the 10/66 Dementia Research Group. Surveys assessed residents aged 65 and older in predefined catchment areas in China (n=2,162), India (n=2,004), and six Latin American countries (Cuba, Dominican Republic, Mexico, Peru, Puerto Rico, Venezuela; n=12,865) between 2004 and 2006 (2007–2010 for Puerto Rico). Risk factor prevalence from these surveys was combined with relative risk estimates from published meta-analyses and adjusted for communality to calculate weighted PAFs.
What was found
The overall weighted PAF for modifiable dementia risk factors was 39.5% (95% CI 37.5–41.6) in China, 41.2% (95% CI 39.1–43.4) in India, and 55.8% (95% CI 54.9–56.7) in the Latin American sample, compared to 35% worldwide. Five risk factors had higher PAFs than global estimates: less childhood education (weighted PAF 10.8% in China, 13.6% in India, 10.9% in Latin America vs 7.5% worldwide), smoking (14.7%, 6.4%, 5.7% vs 5.5%), hypertension (6.4%, 4.0%, 9.3% vs 2.0%), obesity (5.6%, 2.9%, 7.9% vs 0.8%), and diabetes (1.6%, 1.7%, 3.2% vs 1.2%).
Why it matters
Modifiable risk factors account for a substantially higher proportion of dementia cases in low- and middle-income countries than in high-income countries. Public health strategies targeting early-life education, hypertension, smoking, and metabolic health could prevent a large fraction of global dementia burden.
Limits
The analysis used cross-sectional prevalence data that cannot establish causality. Relative risk estimates were borrowed from external meta-analyses predominantly conducted in high-income countries, which may not reflect the true risk relationships in LMICs. Sampling was restricted to specific catchment areas and may not fully represent entire national populations, and surveys were conducted between 2004 and 2010, which may not reflect current prevalence rates.
Cited by
- supports Modifiable lifestyle factors studied by Gill Livingston and colleagues, including poverty, childhood obesity, lower education, smoking, and excessive alcohol use, substantially influence the risk of developing dementia and Alzheimer's disease.