Working alliance and adherence mediate the effect of guidance in a web-based program for participants with mild to moderate depressive symptoms: A secondary mediation analysis

Guided web-based self-help programs for individuals with depressive symptoms have shown to be more effi-cacious than unguided programs. However, research has paid little attention to why guided interventions are superior. The present study investigated whether working alliance and adherence to the program mediated the effect of guidance on depressive symptom outcome.The study is a secondary analysis of a randomized factorial trial. In the trial, 302 adults with mild to moderate depressive symptoms (Patient Health Questionnaire-9 score: 5-14) were randomized to either a guided or an unguided group. All participants received access to a web-based self-help program based on problem-solving therapy. Working alliance with the treatment providers was assessed using an adapted version of the Working Alliance Inventory for Guided Internet Interventions two weeks (early-treatment) and eight weeks (post -treat-ment) after pre-treatment. The primary outcome was depressive symptoms at post-treatment.The total working alliance score was significantly higher for guided participants compared to unguided par-ticipants (at early-treatment: t248.6 = -3.36, p < .001, d = 0.42, at post-treatment: t194.9 = -4.77, p < .001, d = 0.66). The total working alliance score correlated significantly with the change in depressive symptoms for guided (rs = 0.16, 0.34) and unguided participants (rs = 0.26, 0.23). The WAI-I total score statistically mediated the relationship between guidance and outcome (at early-treatment: B = -0.028, at post-treatment: B = -0.053). Furthermore, the subscale tasks (at post-treatment: B = -0.051), the subscale goals (at early-treatment: B = -0.031 and at post-treatment: B = -0.052), and adherence to the program (B = -0.034) mediated the relationship between guidance and outcome. Finally, in a multiple mediation model both early-treatment working alliance and adherence to the program (B = -0.050) mediated the relationship between guidance and outcome.These findings indicate that guidance increases working alliance to treatment providers as early as two weeks after treatment beginning. The alliance predicts outcome and mediates the relationship between guidance and outcome. Participants' agreement with tasks and goals of a program seems to be more important than the bond with treatment providers. Treatment providers might therefore attune web-based programs to the preferences and expectations of participants. In addition to the working alliance, adherence to the program co-mediates the relationship between guidance and outcome.