Human contact in internet-based interventions for depression: A pre-registered replication and meta-analysis of randomized trials

Introduction: Internet-based self-help interventions have shown to be effective in the treatment of depression. Several meta-analyses indicated that human contact has a crucial impact on adherence and outcome. While most research focused on the role of guidance during interventions, a review by Andersson and Johansson (2012) suggested that contact before the intervention too may play an important role. Objective: We investigated the impact of the degree of contact in internet-based interventions (IBIs) for depression on outcome in adults suffering from elevated symptoms of depression. Methods: We conducted a preregistered meta-analysis (www.osf.io/4mzyd) and included trials comparing IBIs for depression against control conditions (treatment as usual [TAU] or waiting list [WL]) in patients with symptoms of unipolar depression searching the databases PsycINFO and Cochrane's Central Register of Controlled Trials (CENTRAL) limited to entries from EMBASE and PubMed. Following Andersson and Johansson (2012), contact before an intervention was defined as having had a diagnostic interview before the IBI, and contact during intervention was defined as having received guidance during the IBI. IBIs were grouped as providing (0) no contact, (1) contact before the IBI, (2) contact during the IBI, or (3) contact both before and during the IBI. The primary outcome was standardized mean difference (SMD) of the IBI and control in depressive symptoms at treatment termination. Secondary outcomes were study dropout and adherence to the IBI. Results: We included 56 eligible trials that randomized 13,335 patients to 75 internet-based intervention conditions and control groups (TAU in 23 trials, WL in 33 trials). In total, 44 trials (78.57 %) were judged to show some concerns or a high risk of bias. Overall heterogeneity was high regarding the primary outcome (I2s