. 2021 Jul 16;10(14):3151. doi: 10.3390/jcm10143151

Table 2.

Results of the analysis with three mediators based on pooled data from three trials (subset of patients with tinnitus at baseline, n = 594, full analysis set); parameter estimate, standard error (SE) and p-value.

Effect	Estimate	SE	p-Value
a1: treatment effect on depression (model 3)	0.7112	0.130	<0.001 ⁽¹⁾
a2: treatment effect on anxiety (model 3)	0.8051	0.148	<0.001 ⁽¹⁾
a3: treatment effect on cognition (model 3)	2.9287	0.219	<0.001 ⁽¹⁾
b1: mediator effect (depression) on tinnitus (model 2)	0.1234	0.035	0.001 ⁽¹⁾
b2: mediator effect (anxiety) on tinnitus (model 2)	0.1031	0.030	0.001 ⁽¹⁾
b3: mediator effect (cognition) on tinnitus (model 2)	0.0912	0.0204	<0.001 ⁽¹⁾
c’: direct effect (model 2)	0.6169	0.120	<0.001 ⁽¹⁾
c: total effect (model 1)	1.0549	0.110	<0.001 ⁽¹⁾
a1 × b1: indirect effect mediated by reduced depression	0.0877	0.030	0.004 ⁽²⁾
a2 × b2: indirect effect mediated by reduced anxiety	0.0830	0.029	0.004 ⁽²⁾
a3 × b3: indirect effect mediated by improved cognition	0.2672	0.063	<0.001 ⁽¹⁾

⁽¹⁾ Two-sided t-test p-value; ⁽²⁾ p-value of Sobel’s test.