Table 2.
Mediation of health behavior change between intervention and health outcomes (18 months, n = 240)
|
T0–T2
Fasting insulin (pmol/L) |
X -- > Y (c path) b,i | ||||||||
| B(SE) | 95%-CI | ||||||||
| Crude analysis | −6.8 (3.4) | −7.2; −6.4 | |||||||
| X -- > Y (c’ path) c,i | X -- > M (a path) d,i | M -- > Y (b path) e,i | Indirect effect (a*b) f | Proportion mediated g | |||||
| B (SE) | 95%-CI | B (SE) | 95%-CI | B (SE) | 95%-CI | B (SE) | 95%-CI h | (a*b)/c | |
| NBS indexa | 4.9 (0.9) | 3.2; 6.7 | −0.5 (0.3) | −1.1; 0.1 | −2.4 (1.4) | −5.3; 0.0 | 0.34 | ||
| MVPAa | 1.6 (0.5) | 0.6; 2.7 | −0.3 (0.5) | −1.2; 0.7 | −0.4 (0.9) | −2.6; 1.0 | - | ||
| Combined | −4.3 (4.2) | −12.7; 4.0 | −2.9 (1.6) | −6.3; −0.1 | 0.40 | ||||
|
T0–T2
Body weight (kg) |
X -- > Y (c path) b,h | ||||||||
| B(SE) | 95%-CI | ||||||||
| Crude analysis | −2.4 (0.6) | −3.5; −1.3 | |||||||
| X -- > Y (c’ path) c,h | X -- > M (a path) d,h | M -- > Y (b path) e,h | Indirect effect (a*b) f | Proportion mediated g | |||||
| B (SE) | 95%-CI | B (SE) | 95%-CI | B (SE) | B (SE) | 95%-CI h | B (SE) | ||
| NBS indexa | 4.9 (0.9) | 3.2; 6.7 | 0.0 (0.04) | −0.1; 0.1 | −0.1 (0.2) | −0.4; 0.3 | - | ||
| MVPAa | 1.6(0.5) | 0.6; 2.7 | −0.2 (0.1) | −0.4; −0.1 | −0.4 (0.2) | −0.9; −0.1 | 0.17 | ||
| Combined | −1.9 (0.6) | −3.1; −0.8 | −0.5 (0.3) | −1.0; 0.0 | 0.20 | ||||
aNBS index defined as a score ranging from 0 to 60, MVPA (# times, 30 min per week)
bC path (total effect): the association between intervention and outcomes (fasting insulin and body weight). All analyses are adjusted for baseline value, sex, recruitment phase, and medication use
cC’ path (direct effect): the association between intervention and outcomes, adjusted for mediators (NBS index and PA)
dA path: association between intervention and NBS index or PA
eB path: association between NBS index or PA and outcomes
fIndirect effect (a*b): the indirect effect of the intervention on outcome through NBS index and/or PA
gProportion effect mediated ((a*b)/c): the proportion of the total effect that was mediated through NBS index and or PA
hStandard error and confidence interval for indirect effects were calculated with bootstrapping (5000 samples)
iAll analyses use linear regression models