Risky drinking and other drug use in adults with chronic conditions in the United States: differential associations by race/ethnicity

Abstract

Co-use of multiple drugs may prolong or increase heavy drinking, even for individuals with health conditions adversely affected by it. Patterns of alcohol and drug use may vary across racial/ethnic groups, with differential implications for health. This study examines racial/ethnic differences in the associations between risky drinking and other drug use in adults with diabetes, hypertension, heart disease, or cancer. Multiple logistic regression modeling, stratified by condition, was performed using a nationally representative sample of adults drawn from the 2015 to 2019 National Survey on Drug and Health. The outcome was risky drinking (consuming more than 7/14 drinks weekly). Other drugs considered were tobacco, marijuana, illicit drugs, and non-medical prescription drugs. Covariates included age, sex, education, income, marital/cohabitation status, health insurance coverage, and self-rated health status. Each drug category was positively associated with risky drinking across all four conditions. Racial/ethnic minority adults were less likely than White adults to engage in risky drinking, with this pattern most consistent for those with hypertension. Other drug use in minority adults (i.e. tobacco and illicit drug use in Black and Hispanic adults, and marijuana and prescription drug use in Asian adults) was associated with disproportionately greater odds of risky drinking compared with White adults. This pattern was more prominent for those with a heart condition, and not found for those with cancer. Future interventions might address co-use of alcohol and other drugs in adults with chronic conditions, with special attention to racial/ethnic minority adults.

Introduction

Chronic health risks associated with heavy drinking are well-documented. Heavy drinking is consistently associated with increased risks for heart disease and cancer (Rehm et al. 2017), the top two causes of death in the United States (US) (Murphy et al. 2021), and for diabetes and hypertension, two common conditions that increase risk for both heart disease and cancer (GBD 2015 Risk Factors Collaborators 2016). Continued heavy drinking post-diagnosis of these conditions may carry a risk for long-term health harms. One study, for example, found patients with Type 2 diabetes who reported heavy alcohol consumption had dose-dependent higher risks of cardiovascular events and all-cause mortality, compared with their counterparts who did not consume alcohol (Blomster et al. 2014). Continued drinking post-diagnosis of head and neck cancer was associated with significantly higher mortality risk (Mayne et al. 2009).

Individuals who use one drug often take other drugs as well. An estimated one in five adults living in the US used alcohol and tobacco (Falk et al. 2006), and next to tobacco, marijuana is the most commonly used drug among drinkers (Substance Abuse and Mental Health Services Administration 2013). Many illicit drug users also drink, for example, with about three-quarters of cocaine users consuming alcohol (Liu et al. 2018). Non-medical use of prescription drugs in college students was significantly associated with binge drinking (Kilmer et al. 2021).

Co-use of alcohol and other drugs may lead to more adverse health outcomes than alcohol use alone. Smoking synergistically interacts with alcohol intake to increase the level of the liver enzyme serum gamma-glutamyltransferase (Breitling et al. 2009), a key predictor of common diseases including diabetes (Meisinger et al. 2005) and coronary heart disease (Meisinger et al. 2006). Co-users of alcohol and tobacco have much greater risk of developing cancers than people who use either alcohol or tobacco alone (Hashibe et al. 2009). The combination of alcohol and cocaine, the latter an independent risk factor for adverse cardiovascular outcomes (Winhusen et al. 2020b), was found to have greater-than-additive effects on heart rate (Pennings et al. 2002). Although findings are mixed (Zhao et al. 2021), there is evidence that marijuana use predicts adverse cardiovascular outcomes (Winhusen et al. 2020a). The use of other drugs makes it difficult to reduce alcohol consumption. For one, smoking initiators and persistent smokers had greater odds of alcohol and marijuana relapse compared with smokers who quit (de Dios et al. 2009). Marijuana use was associated with increased risk of alcohol use disorder (AUD) onset and persistence (Weinberger et al. 2016), and marijuana use in adults with AUD was associated with lower odds of achieving abstinence from heavy drinking compared with non-users (Mojarrad et al. 2014, Subbaraman et al. 2017).

Despite the elevated health risks co-use might incur (in part by prolonging or increasing heavy drinking), there is a paucity of research on the co-use of alcohol and other drugs in adults with a chronic condition that is adversely affected by drinking. We are unaware of studies on this topic using nationally representative adult samples, particularly with a specific focus on race differences. Patterns of alcohol and drug use may vary across racial/ethnic groups. Health risk behaviors including alcohol and other drug use are influenced by the social, cultural, and economic circumstances that frame and constrain them (Cockerham 2000), which may significantly vary across diverse racial/ethnic groups. There is also ample evidence of racial disparities in the burden of these conditions and associated mortality in the US. Prevalence of hypertension, for example, is higher among Black adults than among other racial groups in the US (Virani et al. 2020), and prevalence of diabetes is higher among Hispanic adults than all other groups except American Indians/Alaska Native adults (Centers for Disease Control and Prevention 2020). Heart disease mortality rate was found to be far higher for Black than for White adults (Van Dyke et al. 2018), as was cancer mortality (Siegel et al. 2022). Fundamentally, racial disparities in health have been attributed to stigmatized social status intertwined with lower socioeconomic status affecting many members of racial minority groups (Williams and Sternthal 2010). In addition, greater prevalence of more proximate, modifiable risk factors including obesity and smoking in some populations have also been implicated (Agarwala et al. 2021). Risky drinking and other drug use may be among such factors. A better understanding of the patterns of the associations between risky drinking and other drug use, which may vary across racial groups, might help inform tailored interventions to improve the health of individuals with chronic conditions.

Using a nationally representative sample of US adults with diabetes, hypertension, heart condition, or cancer, this study aimed to examine associations of other drug use (tobacco, marijuana, and illicit drug, and non-medical prescription drug use) with risky drinking (consumption of more than 7/14 drinks weekly for women/men (United States Department of Agriculture 2020), and the moderation of these associations by race/ethnicity. Informed by research that drinking following onset of chronic conditions vary by condition (Kerr et al. 2017), we stratified these analyses by condition.

Methods

Data

A sample of adults aged 18 and older was drawn from the 2015–19 National Survey on Drug and Health (NSDUH). The NSDUH is a nationally representative survey of the US non-institutionalized civilian population including persons with no permanent residence and those living in college dormitories, in-group homes, and on military bases. It uses multistage, nonprobability sampling methods for each of the 50 states and the District of Columbia to select a representative sample of respondents. Adults 18–25 were oversampled in years 2015–19.

Measures

Each health condition was indicated by a dichotomous variable that a health professional had ever told the participant they had diabetes, a heart condition, hypertension, or cancer. Our heavy drinking measure was past month risky drinking (vs. a lower level of drinking or abstinence) of having more than 7/14 drinks weekly for women/men, which exceed current US Dietary guidelines (United States Department of Agriculture 2020). To construct this variable, we first calculated drinking volume using responses for past 30-day drinking frequency, usual quantity (number of drinks on average on days when drank), and binge (4/5 drinks for women/men) frequency. Given respondents’ tendencies to underreport alcohol consumption, we incorporated binge frequency into our calculation of volume by following previous work (Subbaraman et al. 2020). Volume was derived using (frequency days – binge days) × usual quantity + binge days × 6 or 7 drinks (for women and men) for respondents whose usual quantity was fewer than 5/4 drinks (for men/women); and by multiplying frequency days and usual quantity for those whose usual quantity was 5/4 drinks. This volume variable was then converted to average weekly and daily consumption to derive the risky drinking measure. To account for those missing on the drinker status variable and not having information for usual binge days (n = 660, 19.3%), we imputed volume using only frequency and usual quantity.

Measures of other drug use included: tobacco use (any use of tobacco products including cigarettes, smokeless cigarettes, roll-own tobacco, pipe tobacco, or cigar in the past month), marijuana use (any use of marijuana in the past month), illicit drug use (any use of cocaine, crack, heroin, methamphetamine, hallucinogen, or inhalant in the past year), and non-medical prescription drug use (any past-year use of a pain killer, prescription drug, or stimulant in a way not directed by a doctor).

Covariates included demographic characteristics: sex (male vs. female), age (18–34, 35–49, 50–64, 65, and older), race (White, Black, Hispanic/Latinx, Asian, Other), education (less than high school diploma, high school diploma, some college education, four-year college/advanced degree), family income (living in poverty, income up to twice the Federal Poverty Threshold (FPT), greater than twice the FPT), and marital/cohabitation status (married or living with a partner vs. widowed, divorced, separated, or never married). We also adjusted for sexual minority status (being lesbian, gay or bisexual vs. heterosexual), past year health insurance coverage (vs. no coverage), and self-rated health status (excellent, very good, or good vs. fair or poor). Sexual minority status has been positively associated with alcohol and other drug use (McCabe et al. 2009). Health insurance coverage generally is negatively associated with alcohol and drug use (Wang and Xie 2017). Self-rated health status, which is a proxy for overall physical health, was associated with alcohol and drug use (Mauro et al. 2015).

Statistical analysis

Analyses were conducted using STATA version 15.1. We first performed univariate analyses to understand sample characteristics, and bivariate analyses to examine race differences in demographic characteristics, risky drinking and other drug use, and proportions of individuals with each condition. A main effects regression model was then estimated for associations of risky drinking with tobacco, marijuana, illicit drug, and non-medical prescription drug use for individuals with each condition, controlling for age, sex, race/ethnicity, education, income, marital/cohabitation status, health insurance coverage, sexual minority status and self-rated health status. To test moderation by race/ethnicity in these associations, we included interaction terms between race and other drug use. Analyses were stratified by condition. To facilitate interpretation, interactions were graphed for each racial/ethnic minority group, and for each drug use (vs. no use of the drug).

Results

Sample characteristics

In our sample of US adults who had diagnosed diabetes, hypertension, heart disease or cancer, women were somewhat overrepresented in all four racial/ethnic groups, particularly among Black adults. There were significant race differences in all the variables we considered. White adults were older than racial minority adults, with about 78.3% being ages 50 or older compared with 68.9% of Black adults, 63.0% of Hispanic/Latinx (hereafter referred to as “Hispanic”) adults, and 67.6% of Asian adults. Asian adults had the highest education level among all four racial groups, with about 85.1% having some college education or a 4-year college/advanced degree, compared with 65.2% for White, 51.4% for Black, and 42.6% for Hispanic adults. A similar pattern was observed for family income across the four racial/ethnic groups. Asian adults were also most likely and Blacks least likely to be married or cohabitate with a partner. White and Asian adults were most likely, and Hispanic adults least likely, to have health insurance coverage. Asian and White adults reported having better self-rated health than Black and Hispanic adults. The proportion of individuals who had diabetes was higher among Hispanic adults (48.7%) and that of hypertension higher among Black adults (67.1%) than in other groups. The proportion of individuals with heart disease or cancer was higher among Whites than other groups. The proportion of risky drinkers was the highest among Whites (8.8%) and the lowest among Asians (3.2%). Prevalence of tobacco, marijuana, and illicit drug use was higher among Black adults (who had the same prevalence of illicit drug use as Hispanic adults) than in other groups, and the prevalence of non-medical prescription drug use was higher for White and Hispanic adults than for other groups. Asian adults had consistently lower prevalence of other drug use than other groups (See Table 1.)

Associations of risky drinking with other drug use: main and additive interaction effects

Table 2 shows the results from main effects logistic regression modeling. Unadjusted and fully adjusted models showed similar results. In both unadjusted and adjusted models, other drug use was consistently associated with greater odds of risky drinking for adults with each of the conditions, with the only exception being illegal drugs not significant for those with a heart condition in the adjusted model. Being Black, Hispanic, and Asian were generally associated with reduced odds for risky drinking in adjusted models, more consistently for adults with hypertension, followed by those with cancer, and not significant for individuals with diabetes. Being Asian was more consistently associated with reduced odds of risky drinking (significant for those with hypertension, a heart condition, or cancer) and with larger effect sizes than for being Black and being Hispanic. Also in adjusted models, being Black was associated with risky drinking for adults with hypertension or cancer, and being Hispanic was significant only for those with diabetes.

Table 3 shows results from interaction models. As was the case for main effects models, results did not differ greatly between unadjusted and adjusted models. Some associations not significant in unadjusted models became significant in adjusted models (i.e. the interactions between being Black and illicit drug use for adults with diabetes and for those with a heart condition, and between being Asian and non-medical prescription drug use for adults with a heart condition). In fully adjusted interaction models, significant interactions between race/ethnicity and other drug use were found most prominently for adults with a heart condition. These include the following interactions between: being Black and illicit drug use; being Hispanic and tobacco use and illegal drug use; and being Asian and non-medical prescription drug use. Interaction terms between being Black and tobacco use and between being Asian and marijuana use were also significantly associated with risky drinking for adults with hypertension. The interaction between being Black and illicit drug use was significant for adults with diabetes. Counterintuitively, interaction terms between being Asian and illicit drug use and being Black and non-medical prescription drug use were inversely associated with risky drinking. There were no significant race/ethnicity and other drug use interactions for adults with cancer.

Figure 1 visually depicts these interactions. Although Black and Hispanic adults with a heart condition who did not use tobacco had somewhat lower probability of risky drinking than their White counterparts, the probability of risky drinking was higher for Hispanic than for White adults among those who used tobacco (Fig. 1A). Black and Hispanic adults with a heart condition who used an illicit drug had higher probability of risky drinking than their counterparts among White adults (Fig. 1B). For adults with hypertension, although the probability of risky drinking associated with tobacco use was consistently lower for racial minority adults, the difference in probabilities between tobacco smokers and non-smokers was greater for Black adults than for White adults (Fig. 1C), indicating the disproportionately higher probability of risky drinking for Black adults who used tobacco. Figure 1D shows the higher probability of risky drinking for Asian adults with hypertension who used marijuana, over 3-fold of their White counterparts. Finally, and similar to the pattern shown in Fig. 1B, although the probability of risky drinking associated with non-medical prescription drug use for Asian adults with a heart condition was lower than for their White counterparts, the difference in probabilities between marijuana users and non-users was greater for the former than for the latter (Fig. 1E). (For brevity of reporting, the other 11 graphs depicting non-significant interactions are not included in this submission but will be provided upon request.)

Discussion

This study examined associations between other drug use and risky drinking for adults with diabetes, hypertension, heart disease, or cancer and the moderation of these associations by race/ethnicity. Each drug category was associated with greater odds of risky drinking for adults with each condition (with the only exception being the association with illicit drug use, which was not significant for adults with a heart condition). Interactions between racial/ethnic minority status and some drugs were associated with greater odds of risky drinking, with the specific drug category varying by race/ethnicity and health conditions. Specifically, being Black significantly interacted with tobacco use (for adults with hypertension) and illicit drug use (for those with diabetes or hypertension); being Hispanic interacted with tobacco use and illicit drug use (both for those with a heart condition); being Asian interacted with marijuana use and non-medical prescription use (also for those with a heart condition). No race/ethnic interactions were found for individuals with cancer.

The associations between use of another drug and risky drinking in our main effects models are consistent with past research (Falk et al. 2006, Subbaraman and Kerr 2015, Liu et al. 2018, Kilmer et al. 2021) and thus not surprising. As reported in prior studies, the use of tobacco (de Dios et al. 2009) and marijuana (Subbaraman et al. 2017) makes it difficult to reduce alcohol consumption, and cocaine use also facilitates drinking because alcohol reduces withdrawal or undesired effects of cocaine (Macdonald et al. 2015).

Our novel findings on interactions by race/ethnicity uniquely contribute to alcohol and drug disparities literature. In particular, that racial minority adults who use another drug had disproportionately greater odds of risky drinking, compared with White adults, reveal a new layer of alcohol- and drug-related disparities. In addition to more distal factors involving the fundamental conditions of social contexts and experiences that cause health harms for racial minority groups (Williams and Sternthal 2010), racial/ethnic disparities in the burden of chronic conditions have been attributed to greater prevalence of more proximate, modifiable risk factors including obesity and smoking (Agarwala et al. 2021). Risky drinking associated with other drug use may be among such factors. Our findings suggest that alcohol and other drugs can facilitate use of one another to a greater degree for racial minority adults than for White adults, likely causing greater harms for the former’s health. In this respect, co-use of alcohol and other drugs may constitute another mechanism for racial/ethnic health disparities.

Prior work has reported increases in drinking and persistent heavy drinking in adulthood among Black and Hispanic adults compared with decreases in heavy drinking beyond young adulthood among White adults (Mulia et al. 2017). Our findings suggest a clue to this. That is, persistent heavy drinking in adulthood among Black and Hispanic adults may be attributed, at least in part, to higher prevalence of other drug use in these groups than among White adults (Table 2), which may prolong or increase heavy or risky drinking in adulthood.

The different patterns of interactions we found between specific drugs and race/ethnicity—i.e. tobacco and illicit drug use with being Black or Hispanic, and marijuana and non-medical prescription drug use with being Asian—are also worth noting. The significant interactions between tobacco and illicit drug use with being Black or Hispanic may be due, at least in part, to higher prevalence of use of these drugs among Black and Hispanic adults (Table 1). The prevalence of marijuana (and non-medical prescription drug use) among Asian adults with a health condition, however, was lower than among adults in other racial/ethnic groups (Table 1), which is consistent with past research that found low marijuana use prevalence among Asian adults (Keyes et al. 2017). Greater odds of risky drinking for Asian adults associated with marijuana or non-medical prescription drug use thus were unexpected. In our ad hoc analysis, we found prevalence of marijuana use among Asian adult risky drinkers (23.2%) to be higher than its equivalents among Black (22.1%), Hispanic (18.7%), and White (14.2%) adults. There may be unique processes within Asian communities that facilitate co-use of marijuana (and non-medical prescription drug) and alcohol among adults with chronic conditions. There is little information in NSDUH that can shed light on this. Overall, what accounts for the unique race-other drug interactions associated with risky drinking is not clear.

As noted above, some interactions that were not significant in unadjusted models became significant in adjusted models, which include the interaction between being Black and illicit drug use for adults with diabetes or a heart condition, and between being Asian and prescription drug use for adults with a heart condition. This suggests these racial/ethnic differences are not biologically originated but shaped by their lived environments where various sociocultural and health factors—e.g. family income, age (65+), and self-rated health status, all of which were significantly associated with the outcome in our adjusted models (see Supplemental Table 3 for full results)—play significant roles. Future research examining sociocultural processes within each community that explain this could be informative.

As the health belief model suggests, reductions in health risk behaviors are likely to occur to the extent that individuals perceive negative health consequences of the behavior, their susceptibility to the outcome, and benefits of changing behaviors (Carpenter 2010). Therefore, individuals with health conditions that are adversely affected by risky drinking could be more receptive to messages that inform the health harms of such drinking and advise against it. Tailoring to such individuals may be an effective strategy to reduce risky drinking and improve health. In this respect, our specific focus on adults with diabetes, hypertension, heart condition, or cancer in examining associations between risky drinking and other drug use is of important practical significance, as our study findings might help craft educational messages about health harms of risky drinking in a way to resonate with such individuals. That race interactions associated with greater odds of risky drinking were more consistent for adults with a heart condition than for those with other conditions warrants attention. Given that cardiovascular disease is a life-threatening condition and the leading cause of death in the US (Xu et al. 2022), health messaging about the risks associated with alcohol and drug use among individuals with this health profile could be particularly prioritized.

We acknowledge several limitations of this study. First, common to survey data, self-reported health conditions are a limitation to the extent some respondents have undiagnosed conditions they are unaware of. Still, as this study is intended for providing snapshots of alcohol and other drug use in individuals who know about their diagnosed conditions, with the goal of informing interventions targeting them, underreporting of health conditions is not a serious concern. Second, there was limited consideration of other health conditions beyond this study that may have affected alcohol and drug use. NSDUH provides information on other health conditions such as hepatitis, COPD, asthma, HIV/AIDS, and kidney disease, but adjusting only for the conditions included in the NSDUH is not likely to sufficiently address this limitation. Our strategy, instead, was to control for self-rated health status, using it as a proxy for overall physical health that may influence alcohol and other drug use. Third, some of the individuals in our sample may have more than one condition, which we were unable to adjust beyond adjusting for overall health status. Fourth, we were unable to control for subethnicity, which may be particularly important for Asian and Hispanic populations, because of the lack of ethnicity information in the NSDUH public use data. Given the diversity in socioeconomic status, cultural conditions, and alcohol and other drug use within Asian and Hispanic populations (Vaeth et al. 2017), this is a limitation. Lastly, although the NSDUH is the largest data source for these analyses, small statistical power may have hampered our ability to detect significant interactions between race and lower prevalence drug use (i.e. illicit drugs and misused prescription drugs). The large confidence interval for the interaction term between marijuana use and being Asian might be attributed to the small number of Asian risky drinkers who also used marijuana in our sample. The same may be the case for the counter-intuitive results involving being Asian and illicit drug use and being Black and prescription drug use.

Even with these limitations, the current study meaningfully contributes to the health disparities literature concerning drinking and other drug use. Our race-specific findings concerning differential associations between other drug use and risky drinking across racial groups add important nuances to this literature. In turn, these findings have the potential to help inform alcohol and other drug interventions to improve the health of adults with chronic conditions.

Conflict of Interest: There is no commercial or any other conflict of interest for the authors to declare with regard to the manuscript or the subject matter.

Funding

This work was supported by the US National Institute on Alcohol Abuse and Alcoholism (NIAAA) grant P50 AA005595.

Data Availability

Not applicable (publicly available data were used).

References