Table 2.

Effect sizes and predictors identified in IPD meta-analyses of psychotherapies for adult depression.

Study	Contrast a)	Effect Size	Significant Predictors/Moderators b)	Non-significant Predictors/Moderators c)
	IPD meta-analyses
Bower et al., 2013 [11]	Low-intensity CBT vs. CTR	CBT > CTR: SMD = −0.42 (95% CI: −0.55; −0.29)	SPE: Baseline severity	-
Weitz et al., 2015 [12]	CBT vs. ADM	ADM > CBT (on HAM-D: β = −0.88; p = 0.03)	-	SPE/NSP/MOD: Gender MOD: Baseline severity
Furukawa et al., 2017 [14]	CBT vs. pill placebo	CBT > placebo: SMD: −0.22 (95% CI: −0.42; 0.02)	-	SPE: Baseline severity
Karyotaki et al., 2018 [18]	Guided iCBT vs. CTR	Guided iCBT > CTR; OR = 2.49 response; OR = 2.41 remission	SPE: Older age; native-born; baseline severity	SPE: Sex; relationship; education; medication use; anxiety; previous episodes; alcohol problems
Karyotaki et al., 2017 [16]	Unguided iCBT vs. CTR	Unguided iCBT > control; g = 0.27	SPE: None	SPE: Age, sex, education, relation-ship status anxiety, baseline severity
Kuyken et al., 2016 [17]	MBCT for relapse vs. CTR	MBCT > CTR: HR of relapse = 0.69 (95% CI: 0.58; 0.82)	SPE: Baseline severity	SPE: Age, sex, education, relationship status
Driessen et al., 2020 [18]	Dynamic vs. combined treatment	Combined > dynamic therapy; d = 0.26	-	-
Reins et al., 2020 [13]	Internet interventions for subthreshold depression	Internet interventions > control; d = 0.39	SPE: Higher baseline severity; Older age	SPE: Gender; relationship; employment; previous therapy; medication use; anxiety; medical condition; education
	IPD network meta-analyses		Examined moderators/predictors and models
Furukawa et al., 2018 [19]	CBASP for persistent depression	COMB > CBASP: 2.9 (1.3; 4.6) HAMD points COMB > ADM: 2.9 (1.6; 4.3) HAMD points CBASP = ADM: 0.1 (−1.6; 1.7) HAMD points	Examined moderators/predictors: Baseline depression, anxiety, prior medication, childhood neglect, and several interactions Outcomes: Baseline depression, anxiety, prior pharmacotherapy, age, and depression subtypes moderated their relative efficacy. URL: https://kokoro.med.kyoto-u.ac.jp/CBASP/prediction/ (accessed on 25 November 2021)
Karyotaki et al., 2021 [20]	Guided and unguided iCBT	g-iCBT > u-iCBT: −0.8 (−1.4; −0.2) PHQ-9 points g-iCBT > TAU: −1.7 (−2.3; −1.1) PHQ-9 points g-iCBT > WL: −3.3 (−3.9; −2.6) PHQ-9 points u-iCBT > TAU: −0.9 (−1.5; −0.3) PHQ-9 points u-iCBT > WL: −2.5 (−3.2; −1.8) PHQ-9 points	Examined moderators/predictors: Baseline severity, age, sex, educational level, relationship status, employment status, treatment adherence. Outcomes: Baseline severity was the most important moderator. In mild depression differences between unguided and guided iCBT were small; in more severe depression guided iCBT is more effective. URL: https://cinema.ispm.unibe.ch/shinies/iCBT/ (accessed on 25 November 2021)

^a) Extensive definitions of the different types of therapies can be found in Cuijpers et al., 2020. ^b) Predictors defined as characteristics that indicate whether a patient benefits from a treatment or not. Specific predictors (SPEs) indicate whether a specific characteristic predicts outcome of therapy compared to a no-treatment control, while non-specific predictors (NPEs) indicate variables that are related to improvement, regardless of comparison or control group (within-group improvement). Moderators (MODs) are characteristics that indicate which patients benefit more from one treatment compared to another treatment. ^c) Because these studies are based on large sample sizes, non-significant variables are also important because they are probably not associated with the outcome. Abbreviations: ADM: antidepressant medication; CBASP: Cognitive Behavioral Analysis System of Psychotherapy; CBT: cognitive behavioral therapy; CTR: control group; HAM-D: Hamilton Rating scale for depression; HR: hazard ratio; iCBT: internet-based CBT; IPD: individual participant data; MBCT: mindfulness-based CBT; MOD: moderator; NSP: non-specific predictor; OR: odds ratio; PHQ-9: Patient Health Questionnaire-9 items; SMD: standardized mean difference; SPE: specific predictor; TAU: treatment as usual; u-iCBT: unguided iCBT; WL: waiting list.