Evaluation of a few discrete clinical markers may predict categorization of actively symptomatic non-acute schizophrenia patients as treatment resistant or responders: A study by ROC curve analysis and multivariate analyses

Here, we used Receiver Operating Characteristic (ROC) curve analysis to determine whether clinical factors may aid predicting the categorization of schizophrenia patients as Treatment Resistant (TRS) or antipsychotic responsive schizophrenia (ARS). Patients with an established condition of TRS or ARS were assessed for: clinical presentation and course; neurological soft signs (NES); psychopathology by PANSS; cognitive performances; quality of life scale (QLS); functional capacity; social functioning (PSP and SLOF scales). In ROC curve analysis, significance indicated that the Area under curve (AUC) allowed distinguishing between TRS and ARS. Multivariate analyses were additionally used to provide independent predictive analysis. Multiple clinical variables showed significant AUCs. The largest significant AUCs were found for: NES total score; SLOF Area2; QLS subscale; antipsychotic doses. The highest sensitivity was found for NES total score, the highest specificity for previous hospitalizations. The highest Odds Ratio of being included within the TRS category were found for: NES total score (7.5); QLS total score (5.49); and previous hospitalizations (4.76). This same circumscribed group of variables was also found to be predictive of TRS when adopting stepwise logistic regression or discriminant analysis. We concluded that the evaluation of few clinical factors may provide reliable and accurate predictions on whether one schizophrenia patient may be categorized as a TRS.