In reply
The letter by Andrade and Rao provides two criticisms of our article.1 The first concerns use of the BPRS with alcoholdependent subjects. The BPRS measures a broad range of psychopathologic symptoms, including paranoid ideation, psychosis, depression, and anxiety. Because this was the first systematic study in which naltrexone was given to alcoholics, we were interested in general psychopathology rather than in specific symptoms. Also, we had used the BPRS and the Symptom Checklist-90 routinely for several years in our clinic and therefore had extensive experience with obtaining baseline norms in this population.We were not surprised that neither instrument showed group differences. There may be subtle psychological symptoms caused by naltrexone, and symptom-specific instruments such as the Hamilton Depression or Anxiety Scales would be appropriate instruments. We did not intend to test that specific hypothesis in this study.Second, Andrade and Rao state that the Yates correction is required for