Skip to main content

Table 3 Relative contributions of six components to GrimAA at Y20 from QGC and BKMR by subgroups

From: Relative contributions of six lifestyle- and health-related exposures to epigenetic aging: the Coronary Artery Risk Development in Young Adults (CARDIA) Study

Lifestyle components at or by Y20

Weights from QGC

PIPs from BKMR

By smoking status

Ever smokers (N = 285)

Never smokers (N = 459)

Ever smokers (N = 285)

Never smokers (N = 459)

Alcohol consumption

0.361

0.405

0.868

0.429

Diet quality

 − 0.445

 − 0.762

0.471

0.203

Education years

 − 0.555

 − 0.238

0.676

0.086

Physical activity

0.071

0.325

0.135

0.142

Sleep hours

0.007

0.270

0.138

0.119

Smoking

0.559

NA

1.000

NA

Collective association

1.47 (95% CI 0.39, 2.55), p = 0.008

0.13 (95% CI − 0.43, 0.71), p = 0.639

3.08 (95% CrI 1.13, 5.02)

0.47 (95% CrI − 0.38, 1.32)

By sex

Men (N = 391)

Women (N = 353)

Men (N = 391)

Women (N = 353)

Alcohol consumption

0.136

0.093

0.995

0.401

Diet quality

 − 0.224

 − 0.692

0.012

0.617

Education years

 − 0.776

 − 0.267

0.061

0.193

Physical activity

0.032

0.005

0.109

0.016

Sleep hours

0.034

 − 0.036

0.053

0.055

Smoking

0.798

0.907

1.000

1.000

Collective association

6.67 (95% CI 5.43, 7.91), p < 0.001

3.18 (95% CI 1.92, 4.44), p < 0.001

5.53 (95% CrI 3.78, 7.27)

2.52 (95% CrI 0.74, 4.28)

By race

Black participants (N = 304)

White participants (N = 440)

Black participants (N = 304)

White participants (N = 440)

Alcohol consumption

0.168

0.082

0.993

0.292

Diet quality

 − 0.477

 − 0.423

0.032

0.425

Education years

0.034

 − 0.577

0.063

0.821

Physical activity

 − 0.523

0.061

0.001

0.410

Sleep hours

0.018

0.032

0.028

0.217

Smoking

0.780

0.824

1.000

1.000

Collective association

6.03 (95% CI 4.45, 7.61), p < 0.001

4.52 (95% CI 3.46, 5.58), p < 0.001

5.93 (95% CrI 3.24, 8.61)

1.96 (95% CrI 0.11, 3.81)

  1. Models were adjusted for race, sex, body mass index (BMI), and field center; CI confidence interval, CrI credible interval, QGC Quantile-based g-computation, BKMR Bayesian kernel machine regression, PIPs posterior inclusion probabilities. The positive and negative weights from QGC represents the proportion of the effect estimate for each component (sum up to 1 or − 1 for the same direction); The PIP reflects the ranked importance of each component in association with GrimAA. † Change in mean GrimAA per one quartile change of all six components for QGC; change in mean GrimAA when all of the six lifestyle components are fixed at their 75th percentile compared to when the six lifestyle components are at their 25th percentile for BKMR