The basic symptom criterion 'cognitive disturbances” (COGDIS) and ultra-high risk (UHR) criteria are commonly used for the prediction of psychosis.

However, their predictive value has been assessed so far only by survival analyses using one-time baseline ratings and time-to-conversion. Thereby, potentially risk status-informative fluctuations in risk criteria ratings over time remained unaccounted for.

Therefore we studied if and how the predictive value of COGDIS and the main UHR criterion attenuated psychotic symptoms (APS) and their combination might be influenced by their presence across different assessment times.

In a naturalistic 24-month study, 146 patients at risk for 'cognitive-perceptive basic symptoms” were repeatedly examined (monthly assessments until month 6, thereafter 3-monthly) for COGDIS and APS with the Schizophrenia Proneness Instrument, Adult version, and the Structured Interview of Prodromal Syndromes. Joint latent class analysis was applied to identify different patterns of risk criteria over time and to detect the degree of their association with risk for conversion to psychosis.

The final model included 4 classes: no risk criteria, exclusively BS, exclusively APS and the combination of COGDIS and APS. Class-specific trajectories and survival functions were associated with an increased risk for the conversion to psychosis from a mild to an intense degree, demonstrating a superior performance of the combination of BS and APS.

This result reinforces earlier results of a clearly superior psychosis-predictive value of this combination at baseline and shows that its stability over time. Thus, APS and COGDIS should be repeatedly monitored.