Measuring alcohol-related harm: test-retest reliability of a popular measure.

This study assessed the test-retest reliability of a measure of alcohol-related harm commonly used in cross-sectional surveys. Sixty-four respondents of a 1995 telephone survey participated in a second interview 3 to 5 months after the survey. Drinking status and average volume of alcohol consumed proved to be highly reliable. For the lifetime harm scale, correlation was satisfactory, and reliability fell just short of satisfactory agreement (kappa = 0.716). For a score of alcohol-related harm in the past year, poor reliability was shown (kappa = 0.484). Future research must place greater emphasis on objective indicators and on validation of the measures used.