Working-memory effects of prefrontal tDCS across baseline cognitive-severity groups in schizophrenia: A secondary analysis of a randomized double-blind sham-controlled trial

Background Cognitive impairment is common in schizophrenia and remains poorly responsive to pharmacological treatment. We examined working-memory change after prefrontal transcranial direct current stimulation (tDCS) across baseline cognitive-severity groups and tested whether baseline severity modified response. Methods This complete-case secondary analysis included 120 of 173 randomized participants (62 active tDCS, 58 sham) who completed the MATRICS Consensus Cognitive Battery before and after 10 bifrontal tDCS sessions. K-means clustering of seven baseline domain T-scores identified cognitive-severity groups. Treatment effects were assessed using descriptive comparisons and adjusted mixed-effects models. Benjamini–Hochberg correction was applied jointly across 14 domain-by-group hypotheses, with Holm correction as a sensitivity analysis. To address circularity, groups were re-derived without baseline working memory, and moderation was tested using the time × treatment × group interaction. Results The groups differed mainly in overall cognitive severity rather than profile shape. The active-versus-sham difference in working-memory change was 3.68 T-points (95% CI, 1.57–5.79) in the greater-impairment group and 2.59 T-points (95% CI, 0.90–4.28) in the less-severe-impairment group. Working memory survived joint false-discovery-rate correction in both groups, but Holm correction only in the greater-impairment group. No other domain survived joint correction. After reclustering without baseline working memory, the time × treatment × group interaction was not significant (estimate, 0.93 T-points; 95% CI, −1.71 to 3.56; p = .4915). Conclusions Prefrontal tDCS showed a working-memory signal across both cognitive-severity groups, but baseline severity did not significantly modify response. The groups should be interpreted as positions along a cognitive-severity continuum rather than distinct profiles or subtypes.