Brief Physician Advice for Problem Drinkers-Reply

In Reply.  —We appreciate Dr Bradley's comments regarding the differences in hospital days between the intervention and control groups reported in our article. It is important to clarify that the health care utilization data were based on patient self-report. We also collected utilization data from medical records, claims data, and family member corroborative interviews, but we opted to delay presenting this information until the 48-month follow-up of our study is completed in 1998.The issue of multiple comparisons probably is not a factor because the statistical differences between the groups were large, and the total number of outcome variables, based on previous alcohol treatment trials, was 9. Although bias is always a concern, our trial used randomized controlled trial methods, had a large sample size, recruited subjects from community practices, and had high follow-up rates. In addition, the analysis followed intention-to-treat procedures.Bradley also noted the potential bias of patient self-report. Health