Cognitive deterioration in predicting relapse of psychosis

Abstract

A large proportion of patients with psychosis experience the obstacles imposed by relapse, incurring huge economic and psychological burden. Therefore, relapse prevention is an important target in long-term management of psychosis, and accurate identification of individuals who are at increased risks will largely facilitate this process. Only few predictors have been confirmed to predict relapse, such as medication non-adherence and stressful life events.

Because cognitive deficits have been found to precede initial onset, they could also be relevant to predict subsequent psychotic episodes. Given its potential to be objectively measured and prospectively traced over time, and further echoing the recent trend in mobile health technologies that could support timely and remote assessments, cognitive deficits and/or deterioration could contribute useful insights to relapse monitoring. Nevertheless, its role as an early warning sign in relapse prediction remains largely unexplored.

In this study, cognitive deterioration was explored as a main predictor of relapse, alongside some other clinical and psychosocial factors, in a naturalistic observational study with a one-year prospective follow-up design. A total of 120 patients with psychosis who were in full clinical remission for at least six months were included and followed up for one year, or until relapse, whichever earlier. Following baseline assessment, follow-up assessments were scheduled on a monthly basis, where cognitive performance was measured at each assessment with a verbal (Letter Number Sequencing) and a visual (Visual Patterns Test) working memory task, via a mobile App. Cognitive deterioration was defined as worsened performance over a two-month period prior to relapse, or study termination.

At one year, 18 of 110 (16.36%) patients relapsed. Potential factors related to relapse were first identified through univariate between-group comparisons, which were then entered into a binary multiple logistic regression. The final model explained around 60% of variance in relapse, and revealed the independent and statistically significant contributions of three factors, i.e., cognitive deterioration on the verbal working memory task (Letter Number Sequencing), low resilience, as well as medication non-adherence.

This is one of the first studies to establish the predictive role of cognitive deterioration in psychotic relapse. It not only supports the promising direction for further research into cognitive predictors, but also uncovers important clinical significance of cognitive deterioration in relapse monitoring, early relapse identification, as well as suggests the therapeutic potentials of cognition-based training programs that could further translate into favorable clinical outcomes.