A rapid bedside inventory for post-electroconvulsive therapy cognition: Derivation and validation of the electroconvulsive therapy-cognition inventory (ECT-CI).

BACKGROUND

Electroconvulsive therapy (ECT) is effective for treatment-resistant major depressive disorder (MDD), but peri-treatment cognitive effects require rapid, accurate bedside assessment. We developed and prospectively validated the Electroconvulsive Therapy-Cognition Inventory (ECT-CI).

METHODS

Single-centre, prospective, assessor-masked diagnostic-accuracy/psychometric study including adults with MDD initiating first-course ECT (n = 54; assessed ≤48 h pre-ECT and 7-10 days post-ECT) and age-matched healthy controls (n = 54, single assessment). The reference standard was a DSM-5-TR-anchored, blinded clinical determination of clinically significant cognitive impairment (yes/no) rendered independently by two psychiatrists, who were masked to index-test results.Index tests were ECT-CI (primary) and MoCA (comparator). The primary endpoint was post-ECT ROC AUC; secondary endpoints included pre-post responsiveness, administration time, and inter-rater agreement (Lin's CCC). Prospectively registered (ChiCTR2400094414).

FINDINGS: Among 108 participants, ECT-CI declined from 24.99 to 20.07 after ECT (mean change 4.92; 95 % CI 4.17-5.67), whereas MoCA showed no change. Post-ECT discrimination favoured ECT-CI over MoCA (AUC 0.972 [95 % CI 0.928-1.000] vs 0.852 [0.751-0.953]). The operational ECT-CI cut-off (≤21) yielded sensitivity 0.970 and specificity 0.905; MoCA's optimal cut-off (25.5) yielded sensitivity 0.909 and specificity 0.667. ECT-CI required 5.9 ± 2.1 min versus 10.7 ± 3.9 min for MoCA. Inter-rater reliability for ECT-CI was high (Lin's CCC 0.922; 95 % CI 0.810-0.969).

INTERPRETATION: ECT-CI outperformed MoCA for detecting clinically defined post-ECT cognitive impairment, was more responsive to short-term change, and was briefer to administer. An operational threshold of ≤ 21 supports pragmatic bedside use; external multicentre validation is warranted.