Revisiting the relapse predictive validity of prodromal symptoms in schizophrenia.

After recovering from a first episode in schizophrenia, relapse prevention is a major objective and long-term treatment is indicated. Due to empirical findings regarding relapse preventive efficacy of antipsychotics, guidelines recommend maintenance neuroleptic treatment, which however is not feasible for all patients. Both under maintenance treatment and drug withdrawal, doctors and patients try to anticipate a reexacerbation by means of early signs of relapse in order to intervene as early as possible. Prodromal symptoms are supposed to be such early indicators, although empirical evidence is not uniform. In order to prepare a new research program, data from an earlier study were reanalyzed to examine the relapse predictive validity of prodromal symptoms. 339 outpatients observed prospectively for 2 years under different long-term treatment strategies were included in this analysis resulting in a total amount of 5861 observations/visits. Several prognostic parameters were computed regarding relationship between clinical state (relapse yes/no) and the occurrence of prodromal symptoms the immediate visit before. As to the results, none of the 22 single prodromal symptoms exceeded a sensitivity of 40%, with specificities ranging from 69% to 95%. An overall prodrome score led to a sensitivity of 72%, however specificity declined to 38%. The performed logistic regression analyses including other potential predictors (psychopathology, clinical change etc.) yielded other relevant 'early signs'. It is concluded that the relapse predictive validity of prodromal symptoms has to be enhanced e.g. by supplement of other clinical parameters with predictive potential or by closer monitoring.