The current study confirms how early onset of substance use affects the development of adult binge drinking behavior and expands our understanding by quantifying the impact of early onset of substance use on the severity of binge drinking. All three predictors of onset of substance use prior to age 18, including alcohol, smokeless tobacco, and marijuana, were identified as independent predictors of binge drinking and were found statistically significantly associated with the binge drinking to various extent. Onset of alcohol use was the strongest, following smokeless tobacco, and marijuana use the least. The current study aligns with previous findings that the earlier onset of alcohol drinking, the more likely these individuals are to develop binge drinking and other alcohol use related problems [10,24]. However, some previous studies had contradicted findings. Moss et al. (2014)  found a positive association between early marijuana use and later binge drinking, but a surprisingly negative effect of early cigarettes smoking on adulthood binge drinking. The reasons for varying substance use effects are not clear but might be related to that the different distributional networks vary by substances used. Future research focused on how different patterns or sequences of early onset of substance use (e.g., initiation of alcohol use leads to tobacco use and then marijuana use) affect binge drinking may shed light on this question.
Of great importance, the current study supports that the presence of SUD diagnosis in the past year is the strongest predictor of binge drinking. Pilatti et al. (2017)  has shown that adolescent use of all the three substances was strongly associated with SUD diagnoses in the young adulthood. The strong impact of SUD on binge drinking might be an additive or synthetic effect of the onset of experimenting with substances at an early age . Thus, delaying the onset of substance use, including tobacco, alcohol, marijuana, would be an effective strategy for preventing both addiction and binge drinking in adulthood life.
Demographic characteristics including being male and in young adulthood (age 18-34) are also strongly associated with binge drinking. A national report indicated that between 2001-2002 and 2012-2013, high-risk drinking and alcohol use disorder increased across almost all sociodemographic subgroups in the U.S., especially among women, older adults, racial/ethnic minorities, and the socioeconomically disadvantaged populations . Being African American is marginally statistically (p = 0.05) associated with binge drinking, which is in consistent with a previous NSDUH report . The results revealed the importance of greater public health efforts to reduce binge drinking across these demographics.
Socio-economic factors are mostly marginally statistically (p-value in 0.04-0.09) associated with binge drinking, including low house income (<$49,999), relatively good health (excellent or very good), and no health insurance. SPD in poor mental health was also reported to be associated with binge drinking. These findings are not greatly contradicting with previous studies [20,28]. Alcohol consumption is well-known to be causally associated with and complicated with many chronic diseases including cancer, diabetes, heart diseases, and injuries . However, the present study surprisingly found that higher number of chronic diseases was a protective factor of binge drinking, which aligns with a previous finding that the prevalence of binge drinking was lower among older adults with two or more chronic diseases . The reason for this seemingly unjustified finding might be that patients in multiple chronic diseases are in serious health conditions and have to follow doctor’s orders to avoid alcohol consumption .
The analysis provides a unique strength that measures the severity of binge drinking by number of days in the past month as a count outcome. To our knowledge, the study is the first to investigate risk factors associated with binge drinking as a count outcome using a weighted Poisson regression model in a nationally representative sample. Another unique strength of the current study exists in the classification of age at onset of substance use, the findings of which indicate that the influential range of age at early onset varies by substance. Onset of alcohol use at any age prior to 18 was found statistically significantly associated with the binge drinking; whereas only onset of smokeless tobacco uses after 13 and onset of marijuana use after 15 were found statistically significantly related to the binge drinking.
Despite of many strengths that have shown in this section, our study presents a series of limitations. The cross-sectional design impedes an interpretation of causational factors. The nature of NSDUH data is prone to recall bias, social desirability bias, and non-differential misclassification bias due to the collection of information on past behaviors (e.g., past-month binge drinking and number of days of binge drinking in the past month) [14,19]. There is no worldwide consensus of definition of “binge drinking” or “heavy drinking episode”. The definition used in this study was derived from the US academia and federal agency [21,29], and could be different from other reports or publications worldwide. Another noteworthy limitation to the interpretation of our results is that the study failed to take age of first intoxication by alcohol into consideration. While age of onset of alcohol use has been identified as an independent risk factor of binge drinking, previous studies [10,30] suggest that delay to intoxication may be an important determinant of negative alcohol use outcome and should be considered in the modelling. Although the NSDUH includes various substance use information, the study only includes the main three substances which are the most prevalent among binge drinking population. Besides, unfortunately, this study cannot take interpersonal factors, including parental monitoring and relationships between parents and adolescents, as known risk factors into the modelling binge drinking .