Caffeine consumption and onset of alcohol use among early adolescents

Abstract

Preventing or delaying the onset of alcohol use among children and youth is an important public health goal. One possible factor in alcohol use onset among early adolescents is caffeine. The aim of this study was to assess the possible contribution of caffeine to the onset of alcohol use during early adolescence. We used data from the Young Mountaineer Health Study Cohort. Survey data were collected from 1349 (response rate: 80.7%) 6th grade students (mean age at baseline 11.5 years) in 20 middle schools in West Virginia during the fall of 2020, and again approximately 6 months later in spring of 2021. We limited our analyses to students reporting never having used any form of alcohol at baseline. Logistic regression was employed in multivariable analyses and both Odds Ratios and Relative Risks reported. At follow-up, almost 14% of participants reported having consumed alcohol at least once and 57% used caffeine of 100 mg + daily. In multivariable analyses we controlled for social and behavioral variables known to impact tobacco use. Caffeine use was operationalized as a three-level factor: no use, <100 mg per day, and 100 + mg per day, with the latter being the approximate equivalent of the minimum of a typical cup of coffee or can of energy drink. Caffeine use of 100 mg + per day was significantly related to alcohol use at 6-months follow-up (OR: 1.79, RR: 1.56, p = .037). We conclude that caffeine consumption among 11–12-year-old adolescents may be a factor in early onset of alcohol use.

1. Introduction

Studies have shown that early-onset alcohol use among youth increases the potential for alcohol-related harm later in life, including binge drinking (Windle and Zucker, 2010), increased frequency of drunkenness (Kim et al., 2017), and the development of substance use disorder (Moss et al., 2014). Thus, improved understanding of the factors that may contribute to increased risk of early-onset alcohol use among youth represents an important public health priority. An understudied factor is caffeine consumption.

A recent review suggested that as many as 75% of children and adolescents aged 5–17 consume caffeine regularly (Temple, 2019). Once ingested, caffeine is readily distributed throughout the body, exerting a variety of pharmacological actions at diverse sites both centrally and peripherally, particularly by way of competitive blockade of the neuromodulator adenosine (Ferre, 2016). In addition, both caffeine and alcohol influence dopaminergic function, which has the potential to influence the reward value of psychoactive drugs (Dreher et al., 2009). Consequently, a causal connection between the use of these two common habit-forming substances has high biological plausibility.

In recent years, beverages such as energy drinks with caffeine content similar to, or higher than, regular coffee (~100 mg–200 mg per 8–12 oz. dose; James, 1997) have been heavily marketed towards youth (Pomeranz, 2012; Temple, 2019). However, most research into possible caffeine-alcohol relationships has been conducted with samples of older participants who are already regular alcohol drinkers, including older adolescents and young adults who consume energy drinks (Arria et al., 2011; Marczinski et al., 2016) and alcohol mixed with energy drinks (AmED) (Kristjansson et al., 2015). Generally, such studies have shown that alcohol drinkers who consume AmED show greater desire to consume alcohol (Marczinski et al., 2016), are more likely to consume alcohol more often (Linden-Carmichael and Lau-Barraco, 2017), and in larger amounts (Cobb et al., 2015) compared to non-AmED drinkers. Moreover, results from experimental studies with animals suggest a possible causal relationship between higher levels of caffeine consumption and subsequent alcohol use (Haun et al., 2021; Holstein et al., 2021). However, longitudinal studies with human adolescents are lacking.

An exception to the dearth of evidence from relevant human studies is a finding by Kristjansson et al. (2018), which showed a positive relationship between caffeine use at baseline and both alcohol use and drunkenness one year later after controlling for demographic variables and other substance use at baseline. However, that study did not assess the relationship between caffeine consumption and onset of alcohol use, nor did it control for social and behavioral factors known to be relevant (e.g., peer alcohol use, parental support).

The aim of the present study was to assess the longitudinal relationship between caffeine consumption among early adolescents who report a history of no alcohol use at baseline, and subsequent alcohol use approximately 6 months later, while controlling for well-established risk factors for adolescent alcohol use. We expect that caffeine use at baseline will predict greater likelihood of alcohol onset at follow-up.

2. Methods

2.1. Sample and participants

The present analyses are based on the first two waves of survey data from the Young Mountaineer Health Study (YMHS) cohort. Students enrolled in 20 geographically diverse public middle schools in five counties in West Virginia are being followed twice per year from grades 6 through 8. During the baseline assessment from October–December 2020, 1671 students attended school in either face-to-face or hybrid (part in person, part virtual) formats (i.e., no participants attended in virtual-only format) and thus were accessible to the study team. Of those, 1349 (80.7%) completed the study survey at both baseline and follow-up. The follow-up survey was conducted during April and May of 2021 using identical data collection protocols. At baseline, 1078 participants (79.9%) reported never having used any type of alcohol in their lifetime and thus served as the sample for our analyses.

2.2. Procedure

Students responded to a computer-based survey using the Qualtrics software. Data collection was supervised by research staff. Participants were accessed either inside schools or during designated class-room hours at home, depending on accessibility based on state and county mitigation efforts against the then COVID-19 pandemic. Data collection procedures utilized an honest-broker system to link individual data across study waves while securing individual confidentiality. The institutional review board of West Virginia University approved all study protocols (#1903499093A001).

2.3. Measures

2.3.1. Caffeine consumption

The caffeine measure was designed to assess daily consumption from multiple types of beverages. This inventory has been validated in multiple publications (James et al., 2011; Kristjansson et al., 2014, 2018). Respondents were asked, “How many cups/glasses/cans/ or bottles do you usually drink of the following drinks every day?”: “coffee”, “tea”, “caffeinated soda (e.g., cola drinks, Mountain Dew, Dr. Pepper)”, and “energy drinks that contain caffeine (e.g., Red Bull, Monster, Rockstar, Bolt, etc.)”. Response options ranged from 1 = “None” to 7 = “6 glasses/cups/cans/bottles or more”. In addition, participants were asked: “How many caffeine ‘shots’ (e.g., 5-Hour Energy) do you usually have each day?” and scored on the same scale as the previous four caffeine beverage questions. We then proceeded to convert the caffeine measure into an approximate value of mg/day and operationalized with three factors: “0” = no caffeine use, “1” = daily usage of <100 mg, and “2” = daily usage of 100 mg+, equivalent to a minimum of approximately one cup of coffee, an 8 oz. energy drink, or half the amount of caffeine in a popular caffeine shot.

Alcohol use at follow-up:

Alcohol use at follow-up was assessed with the following question: “In your lifetime, how many times have you had a drink of alcohol of any kind, even just a few sips (e.g. beer, wine, spirits, shots)?” Response options ranged from 1 = “Never” to 7 “40 times or more often”. For our analyses, responses were dichotomized as 0 = “Never”, 1 = “1+ times”.

2.3.2. Control variables

2.3.2.1. Demographic covariates.

Participant gender was assessed with the following question: “How do you describe your gender?”: “Boy”, “Girl”, “Gender Non-conforming”, “Other (Please specify)”. The gender non-conforming and other categories were merged, due to low number of participants (n < 20). Thus, gender was recoded as 0 = “Girls” (reference), 1 = “Boys”, 2 = “Other”. Based on responses to the question, “Who lives in your household”, family structure was dichotomized as “Lives with both biological parents” and “Different arrangements”. Given the low numbers of non-white students in WV, race was dichotomized as “White” and “Other”. Relative family financial status was assessed with the question: “How well off financially do you think your family is compared to other families?”, coded as 1 = “Much better off’ to 7 = “Much worse off’. Responses were recoded into 0 = “Similar to peers” (reference), 1 = “Worse than peers”, and 2 = “Better than peers”.

2.3.2.2. Social and behavioral covariates.

For caregiver support, respondents were first asked to report which individual is their primary caregiver, responded to with a 13-category drop-down menu. By employing skip-logic, respondents were then asked to rate the level of access to primary caregiver support using the 5-item CRPBI-30 instrument (Schludermann and Schludermann, 1988). Sample item; “My [primary caregiver] is able to make me feel better when I am upset”. Response options included: 1 = “Not like”, 2 = “somewhat like” and 3= “a lot like” and were collapsed to form a scale. Peer alcohol use was based on the question “How many of your friends drink alcohol?”, and responses ranged from 1 = “None” to 5 = “Almost all”. Response options for perceived access to alcohol, based on the question “How easy or hard would it be for you to get alcohol if you wanted to”, ranged from 1 = “Very difficult” to 4 = “Very easy”. Participant tobacco use was assessed with three questions: “In your lifetime, how many times have you smoked cigarettes (smoked a whole cigarette not just taken a puff)?”, “used e-cigarettes or vaping devices?”, and “used other forms (for example hookah, snuff, chewing tobacco)?” All questions were scored on a scale ranging from 1 = “Never” to 7 = “40 times of more often”. For the purposes of our analyses, we dichotomized responses as 0 = “No tobacco use” and “1= “Any tobacco use”.

2.4. Analyses

Logistic regression was conducted with caffeine consumption at baseline as the main independent variable. Relative risks were also calculated, and both are reported in Table 2. Missing data patterns were analyzed between all predictors and the dependent variable and found to be statistically significant only for family structure (p = .02), which is controlled for in the analyses.

Table 2.

Logistic regression. Unadjusted and adjusted odds ratios (OR), and adjusted relative risk (RR) associated with ever alcohol use at follow-up (controlling for demographic, social, and behavioral factors at baseline) (N = 1078).

Variable	Unadjusted OR	Adjusted OR	Adjusted OR p value	Adjusted OR 95% Wald CL Lower, upper	Adjusted relative risk
Daily caffeine use (ref = no caffeine)
Less than <100 mg	1.21	1.14	0.590	0.52, 2.52	1.11
100 mg or more	1.82	1.79	0.037	0.88, 3.63	1.56
Gender (ref = girls)
Boys	1.68	1.57	0.920	0.99, 2.49	1.40
Other	3.19	2.64	0.268	0.68, 10.28	2.08
Race (ref = other)
White	0.78	1.00	0.988	0.51, 1.95	1.00
Family structure (ref = other)
Both parents	0.82	1.18	0.482	0.75, 1.86	1.15
Tobacco use (ref = no)
Yes	2.84	2.10	0.103	0.86, 5.15	1.58
Perceived access to alcohol	2.44	2.36	<0.0001	1.93, 2.87	1.87
Primary caregiver support	0.91	0.97	0.046	0.90, 1.05	0.98
Perceived family financial status (ref = same)
Worse than peers	0.88	1.10	0.83	0.45, 2.68	1.09
Better than peers	0.85	1.00	0.87	0.63, 1.59	1.00

3. Results

Descriptive statistics for all study variables are summarized in Table 1. At baseline, approximately 28% of participants reported daily caffeine use of <100 mg per day and 57% reported daily consumption of 100 mg + caffeine. At follow-up, 13.9% of participants reported having ever consumed any type of alcohol. As summarized in Table 2, adjusted analyses revealed caffeine consumption of 100 mg + at baseline was positively related to alcohol use at follow-up (OR: 1.79, 95%CI: 0.88–3.63, RR: 1.56, p = .037). However, caffeine consumption of <100 mg per day was not related to alcohol use at follow-up (OR: 1.14, 95%CI: 0.52–2.52, RR: 1.11, p = .590). No demographic covariates were related to alcohol use at follow-up. Perceived access to alcohol at baseline was significantly related to use of alcohol at follow-up (OR: 2.36, 95%CI: 1.93–2.87, RR: 1.87, p > .001).

Table 1.

Descriptive statistics for all study variables (N = 1078).

Categorical variable	N (%)	Alcohol use at follow-up n (row %)
Alcohol use at follow-up	123 (13.9%)	–
Daily caffeine use at baseline
None	160 (14.8%)	13 (9.6%)
<100 mg per day	302 (28.0%)	28 (11.4%)
100+ mg per day	616 (57.1%)	82 (16.2%)
Gender
Girls	551 (53.2%)	39 (10.2%)
Boys	465 (44.9%)	74 (16.1%)
Other	20 (1.9%)	4 (26.7%)
Race
White	932 (86.5%)	104 (13.5%)
Other	146 (13.5%)	19 (16.7%)
Family structure
Lives with both biological parents	553 (51.3%)	60 (12.8%)
Lives in other arrangements	525 (48.7%)	63 (15.1%)
Family financial status
Worse than peers	73 (7.0%)	8 (13.6%)
About the same as peers	425 (40.9%)	52 (15.1%)
Better than peers	541 (52.1%)	59 (13.1%)
Tobacco use at baseline	51 (4.7%)	12 (30.0%)
Continuous variables	Mean (SD)	Alcohol use at follow-up (n = 123) M (SD)
Peer alcohol use at baseline	1.10 (0.40)	1.39 (0.77)
Perceived access to alcohol	1.46 (0.87)	2.24 (1.11)
Primary caregiver support	13.06 (2.65)	12.4 (2.9)

4. Discussion

We analyzed two waves of survey data from a sample of middle-school students. Analyses were limited to participants who reported having never used alcohol at baseline. Among participants with no prior alcohol use, almost 14% reported having consumed alcohol at 6 months follow-up. More than half of participants reported daily consumption of caffeine of 100 mg or more at baseline. Our analyses revealed a positive relationship between daily caffeine consumption of 100 mg + and onset of alcohol use 6 months later with adjusted OR of 1.79 and RR of 1.56. These findings hold despite controlling for several demographic, social, and behavioral variables known to contribute to alcohol use among youth.

By limiting our analyses to participants who reported never having consumed any form of alcohol at baseline and studying their alcohol use at follow-up (approximately 6 months later) while also controlling for multiple covariates, our findings establish caffeine as a possible, and until now largely unobserved, contributor to early onset of alcohol use among early adolescents. Present findings replicate and extend previous results by Kristjansson et al. (2018), which did not include the breadth of covariates modeled in the present study. In addition to previous cross-sectional studies (James et al., 2011; Kristjansson et al., 2015), these findings suggest that caffeine may act as a gateway drug (Kandel, 2002) for alcohol onset among early adolescents, which in turns suggests that access to caffeine should possibly be limited for minors.

Despite a number of strengths, our study also has limitations, including potentially limited generalizability due to participants being solely from West Virginia (albeit from 20 different schools in five counties). In addition, our sample was predominantly white, and the self-reported nature of our data is potentially subject to recall bias. Moreover, although our measure of caffeine is comprehensive compared to most other studies of child and adolescent caffeine use, our calculation of mg/day is an approximation. Further, we did not include several products known to include small amounts of caffeine such as candy, chocolate, and yoghurt.

In conclusion, our findings support the suggestion that early exposure to caffeine is associated with increased risk of early alcohol use. Rather than a single causal mechanism, multiple biological, behavioral, and social influences are likely to be involved. Notably, both caffeine and alcohol influence dopaminergic function, which has the potential to influence the reward value of psychoactive drugs (Dreher et al., 2009). In that context, the recent advent of highly concentrated caffeine products (e.g., caffeine “shots”) commonly marketed directly at youth, should be of particular concern. More broadly, confirmation that early caffeine exposure may promote early onset of alcohol use should give rise to more general concerns, including consideration about limiting caffeine consumption among children and youth.