Predicting neuroleptic response from a combination of multilevel variables in acute schizophrenic patients

A predetermined set of 22 sociodemographic, psychosocial, clinical, neurocognitive and biochemical potential predictor variables was tested in 98 schizophrenic patients admitted for relapse. The patients were treated with neuroleptics, mostly with haloperidol, for 28 d. Ten of the 22 variables correlated significantly with the neuroleptic response. Using stepwise multiple regression analyses, an optimal combination of 5 predictors was found to be in hierarchical order: disturbances of premorbid adjustment, intensity of positive symptoms at admission, family history of schizophrenia, working ability during the year before admission and serum dopamine-beta-hydroxylase. The 5 best predictors explained 29% of outcome variance, and all 22 variables together explained 35%. Such neurological characteristics as neurological soft signs, handedness, abnormal voluntary movements, spontaneous blink rate and cognitive impairment did not predict the treatment response. Several psychopathological, psychosocial and clinical predictors known from the literature could also be confirmed by cross-validation.