To the Editor.
—The trial of brief interventions for heavy drinking, reported by Dr Fleming and colleagues,1 was impressive with respect to a number of methodological issues. However, I have several comments and questions about their report.First, the data presented regarding health care utilization seem potentially misleading. From the discussion, it appears that the investigators evaluated several measures of utilization based on patients' medical records and found no difference in utilization between the 2 groups (intervention and control). They also evaluated other health-related outcome measures, all of which showed no relationship to the intervention. Only data on patient reports of days hospitalized are presented, and presumably these were not statistically significant when analyzed for each follow-up time separately. The significant ϰ2 value for a trend toward increased days hospitalized in the control group might be, in part, an artifact of multiple statistical comparisons. Bias also may have