LAUSR

Abstract Background Cognition is impaired in patient with psychosis and is predictive of functional outcomes. Despite this, cognitive function is not routinely assessed in clinical services in the United Kingdom. Collecting cognitive data for research is also labour-intensive and expensive. Web-based assessments may be a solution for these issues but to date, these have not been utilised in patients with psychosis or other psychiatric disorders. Methods We have developed an online cognitive battery for use in psychosis research (and broader mental health research) in collaboration with The Many Brains Project, website developers, patients and clinicians (Cardiff ONline Cognitive Assessment, CONCA). Tasks were selected to measure the domains outlined by the Measurement and Treatment Research to Improve Cognition in Schizophrenia (MATRICS) initiative. We have undertaken a cross-validation study in those with schizophrenia (N=15), bipolar disorder (N=16), depression (N=15) and healthy controls (N=19) to compare the online tasks with the MATRICS battery. Following validation, we invited participants from the Cardiff Cognition in Schizophrenia Study (CardiffCOGS) and the National Centre for Mental Health (NCMH) to complete CONCA. Results Correlations between CONCA and MATRICS tasks ranged from 0.25 to 0.73 in our validation sample (N=65). A total of 6960 individuals were invited to participate and 1227 consented to take part. There was a better response rate from NCMH participants (who were recruited more recently) compared to those from CardiffCOGS. Online participants recruited from NCMH were more highly educated (W=1171600, p<0.001) and more likely to be professionals (χ2(1)=5.4, p=0.02) than the original NCMH cohort. In CardiffCOGS, online participants were more highly educated than non-responders (W=7786.5, p=0.003). A total of 887 individuals met inclusion criteria for our analyses including 43 participants with schizophrenia, 146 with bipolar disorder, 261 with unipolar depression, 187 controls and 250 participants with other psychiatric disorders. Consistent with studies using offline assessments, participants with schizophrenia were the most severely impaired group (compared to controls: g=1.36, p<0.001), exhibiting greater impairments than participants with depression (g=1.04, p<0.001) and bipolar disorder (g=0.71, p=0.002). Of note, lower performance on the battery was associated with poorer functional outcome as assessed using the World Health Organisation’s Disability Assessment Scale (B=-1.77, SE=0.3, p=5.8 x 10–9). Discussion Web-based cognitive testing is a suitable method for collecting large-scale data in psychiatric populations, although there was some evidence of recruitment bias. The results of the validation and recruitment phases were used to inform selection of the final battery. We consulted with patients and health professionals from a youth psychosis service and NCMH’s patient involvement group to create a user-friendly interface and will continue to work with these groups to develop clinically useful feedback to facilitate patient monitoring in early intervention psychosis services.