Quantification of racial and ethnic disparities in alcohol-related problems in light of different methodological approaches

Abstract

Aim: We applied the Institute of Medicine (IOM) definition of racial and ethnic disparities in healthcare to estimate disparities in alcohol-related problems. This estimation involved adjusting for drinking patterns, gender and age, with observed disparities further explained by socioeconomic status (SES). We compared results of five statistical approaches which use different methods for adjusting covariates.

Design and setting: We conducted analysis of the repeated cross-sectional data from the US National Alcohol Surveys (NAS) from 2000 to 2020, comparing traditional regression, rank-and-replacement, propensity score weighting, G-computation and the double-robust methods.

Participants: 39 239 respondents aged 18 + across five NAS surveys oversampling Black and Hispanic/Latino/a populations.

Measurements: Our primary analysis examined the dichotomous outcomes of the three alcohol problem measures: occurrence of negative consequences, alcohol dependence (using DSM-IV criteria) and alcohol use disorder (AUD, using DSM-5 criteria). The drinking pattern variables encompassed past year total alcohol volume and measures of heavy drinking, including the number of days consuming 12+, 8-11 and 5-7 drinks.

Findings: After adjusting for age, alcohol volume and heavy drinking days, statistically significantly higher prevalence of DSM-IV dependence and DSM-5 AUD were observed for Black and Hispanic men who drank in the past year compared with White men who drank. For instance, the Black-White difference in AUD prevalence ranged from 3.7% (95% confidence interval = 1.1%, 6.2%) to 4.9% (2.1%, 7.8%)-, while the HispanicWhite difference ranged from 2.3% (0.1%, 4.4%) to 3.4% (1.1%, 5.6%), using different adjustment methods. Further adjusting for SES factors only moderately explained the observed disparities. We found consistent results in the estimation of disparities across all five methods.

Conclusions: There appear to be racial and ethnic disparities in alcohol-related problems between Black and Hispanic men in the United States relative to White men after alcohol drinking patterns and age are adjusted. The findings also exhibit overall consistency across the five different methods or measurement applied.

Keywords: G‐computation; alcohol‐related problems; causal inference; propensity score weighting; racial and ethnic disparities; rank‐and‐replace.