Skip to main content

Table 2 Associations between measures of socioeconomic status and subsequent incident dementia (2013–2018) among National Health and Aging Trends Study participants

From: Socioeconomic disparities in six-year incident dementia in a nationally representative cohort of U.S. older adults: an examination of financial resources

 Model 1 (n = 3917)Model 2 (n = 3785)
hazard OR(95% CI)hazard OR(95% CI)
Income to poverty ratio among participants < 500% povertya0.81(0.710.92)0.84(0.740.95)
Income to poverty ratio among participants ≥500% povertya0.99(0.881.12)0.96(0.771.18)
Financial strain1.21(1.101.32)1.20(1.091.31)
Education0.72(0.630.81)0.73(0.650.83)
Professional occupation1.11(0.981.26)1.10(0.971.24)
Age2.03(1.872.22)1.93(1.762.12)
Race/ethnicity
 White (ref.)      
 Black race1.10(1.011.19)1.09(1.001.18)
 Hispanic ethnicity1.18(1.081.28)1.18(1.081.28)
 Other race/ethnicity1.07(0.971.19)1.05(0.941.17)
Female gender0.98(0.891.08)1.02(0.941.12)
Own home0.98(0.881.09)1.00(0.911.11)
Retired1.16(1.041.29)1.17(1.051.30)
Heart disease   1.01(0.911.11)
High blood pressure   0.97(0.871.08)
Diabetes   1.13(1.011.27)
Stroke   1.10(1.011.19)
Pack-years smoking   1.13(1.031.22)
BMI   0.87(0.770.98)
Depressive symptoms   1.12(1.011.24)
  1. Note: Annual classification of dementia is based on NHATS protocol for classifying ‘probable’ dementia based on (1) cognitive test scores among self-responding participants or (2) AD8 screener scores ≥2 among proxy-respondents or (3) report of physician diagnosis of dementia. Analyses restricted to individuals who were classified as dementia-free in 2012. Standardized coefficients were estimated from discrete survival analysis models in Mplus using full information maximum likelihood. Sampling weights were used to represent the population of Medicare beneficiaries aged 66 years and older in 2012. Model 1 adjusted for 2011 age, gender, race/ethnicity, income to poverty ratio, educational training, whether the participant had mainly a professional occupation, and 2012 financial strain, retirement and home ownership. Model 2 additionally adjusted for 2012 cardiovascular health characteristics, including history of heart attack or other heart disease, high blood pressure, diabetes or previous stroke, pack years of cigarette smoking and BMI, and depressive symptoms
  2. a Associations between income to poverty ratio and incident dementia were estimated using piecewise linear regression. So, the slope was allowed to differ for participants < 500% poverty threshold and participants ≥500% poverty threshold