Abstract / Summary
Abstract Cognitive bias modification for interpretation (CBM-I) is scalable through digital delivery, but its effects on anxiety and depression remain uncertain. We systematically reviewed digital CBM-I studies reporting interpretation bias, anxiety, or depression and synthesised dependent effect sizes using three-level random-effects models. 73 trials with 10,870 participants were included. At post-treatment, CBM-I significantly improved interpretation bias ( g = − 0.54, 95% CI − 0.64 to − 0.43), anxiety ( g = − 0.19, 95% CI − 0.27 to − 0.12), and depression ( g = − 0.16, 95% CI − 0.25 to − 0.07). Effects remained significant at follow-up (ranging between 2–52 weeks) with similar magnitude. Clinical status and stimulus modality moderated post-treatment depression, with larger effects in clinical samples and with auditory delivery. Digital CBM-I produces moderate cognitive and modest but significant symptom benefits, supporting further evaluation in clinical populations, but the predominance of high-risk studies, early-stage trials and limited follow-up evidence does not support readiness for routine implementation.