Estimating Four Decades of Adolescent Binge Drinking in the US: A Modified Poisson Approach


Jiaxin Gu, University of British Columbia; Qiang Fu, University of British Columbia

Despite the decades-long fight against underage drinking on federal and state levels, adolescent binge drinking remains a serious public health and safety concern in the United States. Youth that indulge in alcoholism are more likely to exhibit low academic achievement, long-term physical and emotional impairments, potential cognitive deficits, increasing risks of suicide ideation and attempts, risky behaviors, and other substance abuse during adolescence and early adulthood. Underage drinking also has collateral damages, such as alcohol-related traffic crashes and fatalities of passengers and pedestrians. As by far the most popular psychoactive substance across all ages in the United States, alcohol has been widely consumed by adolescents despite the country’s minimum legal drinking age of 21 years old. Existing institutional efforts to combat with underage drinking include the implementation of the Minimum Legal Drinking Age (MLDA) law in the 1980s and the more recent Surgeon General’s Call to Action to Prevent and Reduce Underage Drinking in 2007. These legislative and administrative efforts, together with intervention and prevention programs, may have strengthened public awareness of adolescent binge drinking and curbed alcohol use among American adolescents. For example, 28.4% of individuals aged 18 reported having at least one alcoholic beverage in the past 30 days in 2022, compared to 72.1% in 1978. Also, their prevalence of daily alcohol use in the past 30 days dropped from 5.7% in 1978 to 1.5% in 2022. Yet, key methodological and conceptual issues remain when analyzing adolescent binge drinking. Given the sensitive nature of substance use and recall bias, existing surveys such as the MTF study, the Youth Risk Behavior Surveillance System, the National Longitudinal Study of Adolescent to Adult Health, and Canadian Health Survey on Children and Youth often use grouped and right-censored counts (GRC) to measure frequencies of adolescent risky behaviors. Due to the lack of statistical tools for analyzing GRC counts, researchers previously treated those counts as categories so that they can be analyzed using multinomial or ordered logistic regression, which resulted in a serious loss of information contained in GRC counts. Meanwhile, adolescent drinking behaviors are often unevenly distributed across socio-demographic groups. It is therefore conceptually important to distinguish between those who are at risk of binge drinking and those who are not, and also distinguish between the overall incidence and incidence of those at risk of binge drinking. By compiling 47 waves of national representative data from the Monitoring the Future (MTF) study, we analyzed two types of adolescent problematic drinking behaviors, past-two-week excessive drinking and past-month drunkenness, using the modified Poisson (mixture) approach to grouped and right-censored counts (GRC). Our results show that the overall decrease in incidence rates was attributable to substantial reductions in the risks of excessive drinking (45.99% in 1980 and 10.47% in 2021) and drunkenness (34.92% in 1998 and 13.14% in 2021). However, at-risk adolescents only showed mild reductions in incidence rates in the last four decades. Being female, Black, attending schools in the South or West, having better academic performance, and growing up in intact families significantly protected adolescents against binge drinking. Adolescent binge drinking followed a very gendered pattern, where male adolescents constantly experienced higher risks and drank more heavily than their female peers once at risk. Gender disparities narrowed with time in overall incidence and risks for both binge drinking behaviors and converged for excessive drinking in the more recent years (specifically later stages of the COVID-19 pandemic). We, therefore, suggested that the modified Poisson approach is a necessary and useful tool to separate incidence, risk, and at-risk incidence in epidemiological studies with GRC counts. The alarming high incidence rates of at-risk adolescents, especially those male binge drinkers, warrant further investigation.

Non-presenting authors: Minheng Chen, University of British Columbia; Xin Guo, The University of Queensland

This paper will be presented at the following session: