Virtual reality-based cognitive screening in psychiatry: Investigating optimal cut-offs for cognitive impairment in bipolar disorder

Cognitive impairment is a key treatment target across neuropsychiatric disorders, including bipolar disorder (BD), but standard neuropsychological tools often fail to capture real-world cognitive challenges. Virtual reality (VR) offers a more ecologically valid alternative. This study investigated the optimal cut-off score of the novel Cognition Assessment in Virtual Reality (CAVIR) test for detecting impaired functional cognitive capacity in BD. Cut-off scores on the CAVIR were investigated in a sample of outpatients with BD (primary clinical screening BD sample; <i>n</i> = 68), using the Screen for Cognitive Impairment in Psychiatry (SCIP) as the reference for objective cognitive impairment. Receiver operating characteristic (ROC) analyses were used to identify optimal cut-off scores for detecting cognitive impairment, defined as performance ≥1 SD below the healthy controls (HC) mean (<i>n</i> = 68). Analyses were replicated in an independent mixed sample of patients with mood or schizophrenia spectrum disorders (<i>n</i> = 70) and HCs (<i>n</i> = 70), using a full neuropsychological test battery as reference. Associations between CAVIR, SCIP and interviewer-rated functioning were also explored. In the primary clinical screening BD sample, a CAVIR total score cut-off of ≤64 showed fair sensitivity (74 %) and specificity (71 %) for detecting cognitive impairment with an area under the curve (AUC) of 0.74. This cut-off score proved robust, with the optimal cut-off in the mixed replication sample of ≤66 differing by only two points. CAVIR performance was weakly to moderately associated with SCIP and functioning. The CAVIR demonstrated fair sensitivity and specificity for detecting cognitive impairment, supporting its potential as an ecologically valid screening tool for impaired functional cognitive capacity.