Fig. 2

Mediation analysis with gender (Male or Female) as the predictor, history of depressive symptoms (defined as current status of depressive symptoms within the past 12 months, Yes/No) as the mediator, and the self-reported status of low back pain (LBP) (described as the current status of low back pain, low back pain disorder and any other chronic back defects within the past 12 months, with a binary measure) as the dependent variable. Depicted in the diagram (including β-effect estimates coefficients with their standard error) are all four requirements for a mediation effect which were satisfied: Path a, Path b, and Path c were significant. Path c’ is significantly smaller than Path c. In detail, Path a represents the effect of gender on depressive symptoms. Path b represents the impact of depressive symptoms on low back pain (LBP) controlling for the gender effect. Together Path a and Path b represent the indirect (mediated) effect of gender on LBP through depressive symptoms. Path c represents the total effect of gender on LBP without the effect of the mediator. Path c’ represents the direct effect of gender on LBP and is calculated controlling for the indirect (mediated) effect. The use of bootstrapping statistics indicated that Path c’ was significantly smaller than Path c (P < 0.0001), which supports the fact that depression played a significant role as a mediator. Where;a = path a.. b = path b.. c’ = direct effect, i.e., the effect of gender orientation on the current status of low back pain, mediated by history of depression (adjusted for age, education, marital status, work status and Body Mass Index).. c = effect of sex orientation on the current status of low back pain without the effect of the mediator.. e = β-effect estimates coefficients with standard error.. OR = Odds ratios of the effect.. 95% CI = 95% Confidence Interval. ^⁎⁎p < 0.001.. Δ%=Percentage change