October 1, 1997

Brief Physician Advice for Problem Drinkers-Reply

Author Affiliations

Mercer University School of Medicine Macon, Ga

JAMA. 1997;278(13):1060. doi:10.1001/jama.1997.03550130034021

In Reply.  —The letters submitted in response to the article by Dr Fleming and colleagues1 and my Editorial2 raise important issues. Studies of behavior as complex as problem drinking are essential and difficult. Reliance on self-reported change as a measure of program effectiveness is often necessary; efforts to validate reports by relevant alternate sources should be incorporated whenever possible. Fleming et al conducted a review of medical records and insurance claims data and reported the poor correlation of these data sources with their patient interview data. Choice of patient reports as the primary analysis was appropriate for the design of this study. A more likely confounding variable for their health resource utilization results is the inherent variability of a low-frequency occurrence such as hospitalization in a small, healthy study group. The statistically significant difference in hospital days primarily results from low baseline utilization among the control group and a 4-fold