Table 4

Logistic regression analyses predicting health behaviour outcomes by childhood adversity (financial strain/conflict) with adult socioeconomic disadvantage as the mediating variable (khb method)

TotalDirectIndirectMediated %
OR95% CIOR95% CIOR95% CI
Current smoker
Childhood adversity
 Strain (−), conflict (−)111
 Strain (+), conflict (−)1.321.16 to 1.511.161.02 to 1.331.141.09 to 1.1946.4
 Strain (−), conflict (+)1.961.67 to 2.291.891.62 to 2.211.030.99 to 1.08
 Strain (+), conflict (+)2.111.78 to 2.501.841.55 to 2.191.141.10 to 1.1918.1
Binge drinking—daily/weekly/monthly
Childhood adversity
 Strain (−), conflict (−)111
 Strain (+), conflict (−)0.880.77 to 1.021.221.04 to 1.431.160.96 to 1.40
 Strain (−), conflict (+)1.221.04 to 1.431.231.05 to 1.440.990.98 to 1.00
 Strain (+), conflict (+)1.160.96 to 1.401.200.997 to 1.450.970.95 to 0.98
No sport or physical activity within past 7 days
Childhood adversity
 Strain (−), conflict (−)111
 Strain (+), conflict (−)1.281.13 to 1.461.130.996 to 1.291.131.09 to 1.1849.5
 Strain (−), conflict (+)1.140.95 to 1.371.110.92 to 1.331.030.995 to 1.07
 Strain (+), conflict (+)1.201.00 to 1.451.060.88 to 1.271.131.09 to 1.1867.9
Poor diet—fruit and vegetable consumption never/less than weekly
Childhood adversity
 Strain (−), conflict (−)111
 Strain (+), conflict (−)1.651.04 to 2.631.370.86 to 2.191.211.11 to 1.3137.2
 Strain (−), conflict (+)3.031.88 to 4.872.881.79 to 4.641.050.99 to 1.11
 Strain (+), conflict (+)2.591.58 to 4.252.141.30 to 3.511.211.11 to 1.3220.3
Sedentary behaviour—>3 hours of TV watching per day on average
Childhood adversity
 Strain (−), conflict (−)111
 Strain (+), conflict (−)1.261.10 to 1.451.070.93 to 1.241.171.11 to 1.2369.0
 Strain (−), conflict (+)1.110.91 to 1.361.070.87 to 1.301.040.99 to 1.09
 Strain (+), conflict (+)1.511.23 to 1.861.281.05 to 1.581.181.12 to 1.2439.6
  • Note on the khb method: the total effect is the association between the predictor and outcome (not adjusted for the mediator), the direct effect is the association between the predictor and outcome (adjusted for the mediator) and the indirect effect is the difference between the two in terms of log odds. The mediated percentage is the proportion of the total effect that can be explained by the mediator (indirect effect/total effect).

  • Results in bold are statistically significant (p<0.05). All models adjusted for age, gender, parental occupation and education and country. Mediation analysis was only performed when the total effect was significant.

  • khb, Karlson Holm Breen.