Outcome studies of brief alcohol intervention in general practice: the problem of lost subjects.

AIMS: To identify the attrition rate of eligible subjects from the general practice brief intervention studies, reasons for attrition, and the potential bias arising from lost subjects.
DESIGN: Review of all published trials of brief intervention for excessive drinkers in primary care settings.
FINDINGS: The attrition rate of eligible subjects from the general practice brief intervention studies ranges from 44.3 to 83.2% (mean 70.6%). The potential bias introduced by the characteristics of subjects available and not available for research is not adequately addressed. Where there is evidence, subjects unavailable for study or those lost to follow-up usually show different characteristics (e.g. younger, heavier drinkers, less educated) from those completing the study.
CONCLUSIONS: Study populations in general practice-based brief alcohol interventions may have been those most susceptible to intervention. This suggests caution is appropriate in generalizing from brief intervention study results to routine primary care.