Skip to main content

Table 4 Single and multiple multilevel mediator models for the association between intervention and change in distress (K10) controlling for age, home ownership and employment

From: Analysis of the psychological impact of a vascular risk factor intervention: results from a cluster randomized controlled trial in Australian general practice

  Single –mediator model Multiple-mediator model
αβ (SE) 95% CI of αβ Z (P) αβ (SE) 95% CI of αβ Z (P)
Change in:       
Diet score -0.137 (0.075) -0.284,0.010 -1.833 (0.066 ) -0.172 (0.089) -0.346,0.003 -1.930 (0.053)
BMI 0.014 (0.024) -0.034,0.061 0.568 (0.570) -0.008 (0.019) -0.045,0.029 -0.405 (0.686)
Physical activity score -0.007 (0.018) -0.042,0.028 -0.388 (0.698) -0.013 (0.025) -0.063,0.037 -0.504 (0.614)
Alcohol score -0.027 (0.035) -0.096,0.043 -0.749 (0.454) -0.019 (0.031) -0.079,0.041 -0.623 (0.534)
  1. Notes: Multilevel regression models were adjusted for age, home ownership, employment and within-practice clustering effects.
  2. α estimate of intervention effect on change score of behavioural factors.
  3. β estimate of the independent effect of the change mediator score on change K10 score (controlling for intervention).
  4. αβ (mediated effect) product-of-coefficient estimate.
  5. SE standard error.
  6. 95% CI of αβ 95% confidence interval of the mediated effect.
  7. z standard deviate associated with mediated effect (used for significance testing).