Risk function analysis of disparities in alcohol-related consequences and dependence among Blacks, Whites and Hispanics in the 2000 and 2005 National Alcohol Surveys

The role of drinking pattern in studying alcohol-related problems and dependence symptoms has long been prominent in considering disparities of such problems among different ethnic groups. However, prior assessments of drinking pattern used in predicting risk may have biases varying by ethnic group. This study incorporates recent improvements in consumption measures, adjusting alcohol content in "drinks" from methodological studies of home and on-premise drink pours. It uses data from the 2000 and 2005 National Alcohol Surveys of US adults, including large ethnic oversamples. Dependent measures were 2+ of 15 alcohol-related consequences and, separately, 3+ of 17 DSM-IV dependence symptoms. CHAID (CHi-squared Automatic Interaction Detection), a non-parametric non-linear method of partitioning risk groups, was utilized to examine the effects of both overall intake (volume) and pattern of drinking (frequency of 5+, 8+, 12+ drinks in a day). Risk increased at higher, empirically distinct, volume levels with cut points at 1 drink/week, 1 drink/day, 2 drinks/day, and 4 drinks/day for men (1 drink/month, 1 drink/week, 1 drink/day, 2 drinks/day for women). Within each volume level, frequency of 5+, 8+ or 12+ further partitioned risk, with higher thresholds (e.g., 8+, 12+) differentiating risk at higher volumes. Differences among ethnic groups were examined within the gender-specific risk groups. This analysis strategy addresses both drink size/strength and consumption venue biases, which vary across ethnic groups. Results provide the strongest evidence to date to help inform ethnic disparities observed earlier in risk curves that link specific drinking patterns with alcohol-related consequences or alcohol dependence symptoms.