Impact of treatment as usual on efficacy estimates of face-to-face and internet-based psychotherapy for depression: A pre-registered systematic review and meta-analysis of randomized trials
Background: Treatment as usual (TAU) is the most frequently used control group in randomised trials of psychotherapy for depression and meta-analyses that summarise these trials. Concerns have been raised over imprecise and biased efficacy estimates because of substantial variability in the treatments provided in TAU. We set out to investigate the impact of the control group intensity (i.e., quantitative and qualitative aspects of treatments in control groups) on results of trials of psychotherapy for depression and confounding due to systematic differences in this intensity in trials of face-to-face (F2F) and internet-based (INT) psychotherapy.Methods: We conducted a pre-registered systematic review and meta-analysis. We searched PsycINFO for systematic reviews of psychotherapy for depression from Jan 1, 2015 to Jan 3, 2020 and retrieved all references of trials included in these systematic reviews. We searched Cochrane’s Central Register of Controlled Trials and PsycINFO for randomised trials from Jan 1, 2018 to Jan 3, 2020. We included trials that compared (individual or group) F2F or (self-guided or guided) INT with TAU or waiting list (WL) in the acute treatment of patients with symptoms of unipolar depression. We excluded trials published before 2000. All information was extracted by two independent researchers. Six dichotomous indicators were used to create a score that expressed the intensity of TAU and WL. Primary outcome: Standardised mean differences (SMD) of treatment and control groups in depressive symptoms at treatment termination. SMDs were extracted from trial reports. A protocol was pre-registered with the Center for Open Science (www.osf.io/4mzyd).Findings: We included 89 eligible trials that randomised 14,474 patients to 113 eligible psychotherapy arms (8,284 patients) and 89 control arms (6,190 patients). TAU was used as control group in 42 trials (47.19%) and WL in 47 trials (52.81%). F2F was investigated in 37 trials (42.05%) and INT in 51 trials (57.95%). Intensity of control groups did not differ between F2F trials (M = 0.86, SD = 1.42) and INT trials (M = 0.39, SD = 0.63, p = .255). Intensity of control group was a significant predictor of trial results in the main analysis (one-sided p = .042), in pre-registered sensitivity analyses, and in all further exploratory analyses. Heterogeneity estimates were high (I2 = 69.30% to 95.37%). The final analysis found psychotherapy effects to be SMD = -0.287 (95% confidence interval [CI] -0.483 to -0.091, one-sided p = .002, I2 = 76.13%) smaller in 24 trials with higher intensity TAU compared to 16 trials with lower intensity TAU. Interpretation: Our findings suggest that the variability of treatments in TAU influences the results of trials on psychotherapy for depression and is likely to introduce imprecision into meta-analyses of these trials. Source of funding: None