Addictive Behaviors 45 (2015) 252–258

Contents lists available at ScienceDirect

Addictive Behaviors

Young adults who mix alcohol with energy drinks: Typology of risk-taking behaviour Amy Peacock ⁎, Raimondo Bruno School of Medicine (Psychology), University of Tasmania, Private Bag 30, Hobart, Tasmania 7001, Australia

H I G H L I G H T S • • • • •

Typologies of alcohol and energy drink (AmED) risk-taking behaviour were assessed. Low risk-taking, disinhibited intake, and high-risk taking groups were identified. High risk-taking group had elevated risk-taking propensity and alcohol problems. AmED consumers have different patterns of risk-taking behaviour post-consumption. Targeted harm minimisation is warranted for a significant minority.

a r t i c l e

i n f o

Available online 21 February 2015 Keywords: Alcohol Energy drink Caffeine Risk Latent class analysis Risk behaviour

a b s t r a c t Introduction: Contrary to predictions, several studies have shown that people who consume alcohol mixed with energy drinks (AmED) display low odds of risk-taking during AmED versus alcohol drinking sessions. However, these results are based on treating AmED consumers as a homogeneous group. The aim of the present study was to determine typologies of AmED risk-taking behaviour amongst consumers, as well as identifying correlates of AmED risk-taking class membership. Methods: AmED consumers (N = 403) completed an online survey where they reported whether they had engaged in risk behaviours in the preceding six months during AmED and alcohol drinking sessions. Latent class models were estimated based on AmED risk-taking data; univariate multinomial logistic regression was conducted to determine correlates of class membership. Results: A 3-class model was selected based on fit and parsimony, identifying: 1) Low risk-taking consumers (38%): low probability of any AmED risk behaviours; 2) disinhibited intake consumers (48%): high probability of drinking and spending more than intended; and 3) high risk-taking consumers (14%): high probability of most AmED risk behaviours assessed. The latter two groups had significantly higher odds of being male and reporting hazardous alcohol use, more frequent AmED use, greater alcohol and ED intake in AmED sessions, and higher trait impulsivity scores. The latter two groups also reported significantly greater odds of risk-taking behaviours regardless of whether consuming alcohol only or AmED. Conclusions: AmED consumers are not a homogeneous group in regard to their risk-taking behaviours postconsumption. High likelihood of risk-taking behaviour in AmED sessions, as well as elevated risk-taking in alcohol drinking sessions, highlights the need for targeted harm minimisation policies and programmes for a significant minority of consumers. © 2015 Elsevier Ltd. All rights reserved.

1. Introduction With reduction of harmful alcohol consumption a global priority (World Health Organisation, 2010), elevation in hazardous alcohol use amongst young people consuming alcohol mixed with energy drink (AmED) (Patrick, Evans-Polce, & Maggs, 2014) warrants considerable ⁎ Corresponding author at: School of Medicine (Psychology), University of Tasmania, Private Bag 30, Hobart 7001, Australia. Tel.: +61 3 6226 7458; fax: +61 3 6226 2883. E-mail address: [email protected] (A. Peacock).

http://dx.doi.org/10.1016/j.addbeh.2015.02.012 0306-4603/© 2015 Elsevier Ltd. All rights reserved.

concern. AmED use is defined as the co-consumption of alcohol with energy drinks (EDs; caffeinated ‘performance-enhancing’ beverages) packaged pre-mixed, hand-mixed in one beverage, or ingested separately in a drinking session (Peacock, Bruno, & Martin, 2012). A growing body of research shows that people who consume AmED are generally male and younger, and self-report greater and more frequent alcohol use, as well as increased odds of sexual risk-taking, driving risk-taking, illicit drug use, and being physically hurt, injured, or requiring medical treatment, compared to people who consume alcohol (Peacock, Pennay, Droste, Bruno, & Lubman, 2013). Consequently, suggestions

A. Peacock, R. Bruno / Addictive Behaviors 45 (2015) 252–258

that AmED consumption may increase the risk of experiencing alcoholrelated harms by altering the nature of intoxication (Ferreira, de Mello, Pompeia, & de Souza-Formigoni, 2006) are troubling considering the elevated risk-taking propensity of this group. It is theorised that the stimulant effects of the ED mask the sedative effects of alcohol, leading to a reduced sensation of intoxication relative to when consuming alcohol alone (Ferreira et al., 2006). This misperception of intoxication is argued to elevate the likelihood of risk-taking behaviour, as consumers are thought to underestimate their degree of behavioural impairment. In direct contrast with the hypothesis of AmED-induced elevated risk-taking, two studies comparing retrospective self-reported behaviour after AmED versus alcohol consumption in the same individuals showed lower odds of risk-taking in AmED versus alcohol drinking sessions (Peacock et al., 2012; de Haan, de Haan, van der Palen, Olivier, & Verster, 2012; Peacock, Bruno, & Martin, 2013). However, in both studies AmED consumers were treated as a homogeneous group, with no consideration as to whether individual characteristics of the consumer might interact with the pharmacological effects of the drink to lead to differing engagement in risk-taking after AmED. Previous research has shown moderate drinkers with a high proportion of AmED use experienced significantly more alcohol-related consequences than moderate drinkers with a low proportion of AmED use, suggesting that the experience of negative behavioural outcomes may be dependent upon the frequency of AmED consumption (Mallett, Marzell, Scaglione, Hultgren, & Turrisi, 2014). Despite this, no person-centred approach has been undertaken to determine whether there are different patterns of AmED-related risk-taking across the group following consumption of the mixed beverage. Peak medical bodies (e.g., Australian Medical Association; Australian Medical Association, 2010), regulatory bodies (e.g., United States Food and Drug Administration; United States Food and Drug Administration, 2010), researchers (Pennay & Lubman, 2012), and policy-makers (Department of Health and Ageing, 2011) have called for public health reform to address the potential harms of AmED use, with calls to restrict supply through legislation. Identifying consumers with the greatest likelihood of behavioural harms post-AmED consumption, and establishing the correlates of their at-risk status, would elucidate the role of person factors in the use experience; information essential to determine the appropriate target for harm minimisation approaches. Furthermore, seeing whether engagement in risk-taking in alcohol drinking sessions reflects this classification could clarify whether AmED consumers display differential engagement in harmful activities across the two drinking types. Consequently, the aims of the current study were to: i) determine the typology of AmED-related risk-taking behaviour amongst consumers, ii) compare AmED risk-taking profiles based on demographics, alcohol and ED use, and trait impulsivity to determine the correlates of behaviour post-consumption, and iii) determine risktaking behaviour in alcohol drinking sessions according to AmED risktaking typology.

253

Participants with 50% or more responses missing (n = 3) or who reported an international residential status (n = 9) were excluded. As such, the full sample comprised 1101 Australian males and females. Of this group, 421 were identified as AmED users, as they reported: (i) consuming alcohol and EDs in the same drinking session in the preceding six months, and (ii) typically consuming the two constituents simultaneously (i.e., mixed within a single beverage) rather than successively (i.e., as separate beverages within the one drinking session). As the current analyses were restricted to AmED users falling within the target ED demographic (18–35 years), all references to the sample henceforth will refer to AmED users aged 18–35 (N = 403). 2.2. Measures 2.2.1. AmED risk behaviour Participants were asked to report using a dichotomous response format (yes, no) whether they had engaged in 26 risk-behaviours in the preceding six months during: i) AmED drinking sessions and: ii) alcohol drinking sessions (for details of item selection, see Peacock et al., 2012). AmED drinking sessions were defined for participants as occasions when: (i) alcohol and EDs were consumed in the same drinking session simultaneously (i.e., mixed within a single beverage) or successively (i.e., as separate beverages within the one drinking session). Alcohol drinking sessions were defined as occasions when alcohol was consumed and EDs were not co-ingested (within the same beverage or as a separate beverage within the drinking session). Risk behaviours selected represented several theme areas: licit and illicit drug use (e.g., ‘I drank more alcohol than I planned to’), sexual practices (e.g., ‘I had sex with someone I had only recently met’), motor vehicle behaviour (e.g., ‘I did not wear a seatbelt whilst I/someone else was driving a vehicle’), financial outcomes (e.g., ‘I gambled’), aggressive behaviour (e.g., ‘I grabbed, pushed, slapped, punched and/or shoved someone’), mental and physical distress, injury, or harm (e.g., ‘I acted in a way that resulted in me experiencing humiliation or embarrassment’), and other antisocial behaviour (e.g., ‘I was asked to leave or kicked out of a club/bar/ pub’). For the present study, only data relating to risk-taking in AmED sessions was used to determine typologies of risk-taking; low prevalence items were excluded due to low statistical power for analyses, resulting in 10 risk-taking items (see Table 1). 2.2.2. Correlates of risk-taking behaviours Correlates were selected on the basis of being associated with ED and AmED use in past research (de Haan et al., 2012; Brache & Stockwell, 2011; Attila & Cakir, 2011), and included demographics, alcohol and ED consumption patterns, and trait impulsivity. 2.2.2.1. Demographics. Demographic outcomes included sex, age, employment status and education status. To create binary categorical variables, employment was coded as ‘employed’ (full-time, part-time,

2. Materials and method 2.1. Participants Between May and July 2011, 1113 participants aged 18 years or older completed a self-administered online 30-minute survey on independent and combined alcohol and ED consumption patterns. Participants were recruited via posters displayed in the greater Hobart (Tasmania, Australia) area in cafes, bars, nightclubs, and university campuses, as well as media reports and posts on internet forums and social networking sites. Participants were invited to complete the survey regardless of their history of alcohol or ED use. After submitting their responses, participants could redirect to a secure webpage and enter a prize draw to win an Apple iPad 2. The project was granted ethics approval by the Human Research Ethics (Tasmania) Network.

Table 1 Percentage of AmED consumers (N = 403) engaging in risk-taking behaviours in AmED drinking sessions. Risk behaviour

N

% (95% CI)

Smoked cigarettes Drank more alcohol than planned Used illegal drugs Had sex with someone recently met Did not use contraception Spent more money than planned Verbally fought Acted in a way that resulted in me experiencing guilt Acted in a way that resulted in me experiencing humiliation Physically hurt or injured

398 393 395 393 392 397 396 398 396 394

33 (29–38) 62 (57–67) 15 (12–19) 19 (15–23) 16 (13–20) 59 (54–63) 17 (14–21) 26 (22–30) 30 (26–35) 15 (12–19)

Note. 95% CI: 95% confidence interval.

254

A. Peacock, R. Bruno / Addictive Behaviors 45 (2015) 252–258

or casual employment) versus ‘unemployed’ and education was coded as ‘university study’ (currently studying or completed study) versus ‘no university study’. 2.2.2.2. Alcohol and ED use. Participants completed the Alcohol Use Disorders Identification Test (AUDIT; Babor, Higgins-Biddle, Saunders, & Monteiro, 2001), a measure of hazardous alcohol consumption; score range was 0–40 with a higher score indicative of greater current alcohol problems (Babor et al., 2001). In regard to AmED use, participants were asked to indicate: the typical frequency of AmED consumption (less than monthly, 2–4 times per month, 2 or more times per week), and the typical quantity of standard alcoholic drinks (1 standard drink = 10 g alcohol) and standard EDs (one standard ED = 250 mL ED containing approximately 80 mg caffeine) consumed in an AmED drinking session in the past six months. As AmED frequency was typically low (77% of the sample used AmED on a monthly or less frequent basis), this variable was recoded as frequent use (fortnightly or more frequently) versus non-frequent use (monthly or less frequently). 2.2.2.3. Trait impulsivity. Trait impulsivity was measured using the impulsivity subscale from the I7 Impulsiveness Questionnaire (Eysenck, Pearson, Easting, & Allsopp, 1985), a 19-item scale (e.g., ‘Do you often do things on the spur of the moment?’) with a yes/no response format. Scores ranged from 0 to 19, with higher scores indicative of greater rash impulsiveness. 2.3. Analyses Latent class models (one to six classes) were estimated using the 10 risk-taking variables (related to AmED drinking sessions only) and the fit of each model was compared using MPlus program version 7. Following protocols adopted in past published latent class analyses (LCA) (Ramo, Grov, Delucchi, Kelly, & Parsons, 2010), three criteria were used to assess model fit. The first criterion, the Lo–Mendell–Rubin adjusted log-likelihood ratio test (LMR-ALRT) statistic (Lo, Mendell, & Rubin, 2001), was used to compare fit of a k class model with a k − 1 class model, with a low p value (b.050) indicating that the latter should be rejected in favour of the model with one additional class. The second and third criteria adopted were the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (sample-size adjusted; ssaBIC); these indices balance likelihood and model fit, with lower values indicating better model fit. The third criterion, the entropy value, was an index of classification accuracy of the given classes, with higher values (range 0.0–1.0) indicating better differentiation of individuals into classes. Correlates of latent class membership were analysed using univariate multinomial logistic regression conducted in SPSS Statistics Version 21 (IBM, Somers, NY). Descriptive statistics comprised percentages for categorical data and the median for continuous data due to significant positive skew and/or kurtosis. Risk-taking behaviour in alcohol sessions

according to group memberships was analysed using univariate multinomial logistic regression. Across all analyses, significance levels were maintained at p b .050.

3. Results 3.1. Sample characteristics and overall rates of AmED risk-taking The sample was predominantly female (61%), with a median age of 22 years (interquartile range (IQR) = 20–25 years). Participants were relatively well-educated: the majority of the sample (80%) had completed or were currently completing university study and were currently employed (84%). The mean AUDIT score was 12 (SD = 6.0, IQR = 8–16), with 30% scoring ≥16, the cut-off indicative of a high level of alcohol problems and need for brief counselling and monitoring by a health professional (Babor et al., 2001). AmED beverages were typically ingested infrequently (77%), with participants consuming a median of 6 (IQR = 3–10) standard alcoholic drinks and 2 (IQR = 1–3) standard 250 mL EDs in a typical AmED sessions in the last six months; 33% typically consumed EDs in excess of the Australian recommended daily intake guidelines (i.e., more than two standard 250 mL EDs; Food Standards Australia and New Zealand, 2009) in AmED sessions. The median I7 trait impulsivity score for the sample was 7 (IQR = 4– 10), comparable to normative scores for males and females aged 20 to 29 (7.9 and 9.0, respectively) (Eysenck et al., 1985). With the exception of drinking more than intended and spending more than intended (reported by approximately three-fifths of the sample), one-third or less of the sample reported engaging in each of the assessed risk behaviours in AmED sessions in the preceding six months (Table 1).

3.2. Model selection Model fit statistics for the 1 to 6 class solutions for risk-taking in AmED sessions are displayed in Table 2. The fit statistics did not unequivocally support any one solution: (i) BIC was lowest for the 3class model, (ii) ssaBIC was lowest for the 5-class solution and the chisquare test was significant, indicating improved model fit over the 4class solution, and (iii) AIC was lowest, and entropy was highest, for the 6-class solution. The next step in these instances when selecting a model is to examine the classes of competing models for a parsimonious, meaningful and interpretable structure. Examination of the 4-, 5- and 6class solutions did not indicate the existence of an additional, substantive class over the more parsimonious 3-class solution; in these instances, there were one or two substantial classes, with several smaller classes with little difference in response probability, suggesting over-splitting of the sample. The 3-class solution was selected over the 2-class solution, as it splits the smaller cluster evident in the 2-class solution to reveal three substantial classes with distinctive profiles of behaviour.

Table 2 Latent class fit statistics for AmED risk behaviour models with 1 to 6 classes. Model

1 Class 2 Class 3 Class 4 Class 5 Class 6 Class

AIC

4273 3789 3729 3703 3687 3683

BIC

4313 3873 3857 3875 3903 3943

ssaBIC

4281 3806 3756 3739 3731 3737

LMR-ALRT

– 498.435 80.273 47.268 38.026 25.205

LMR-ALRT p value

– b.001 .116 .131 .080 .262

Entropy

– 0.773 0.734 0.742 0.791 0.803

Percentage in each class Class 1

Class 2

Class 3

Class 4

Class 5

Class 6

100 45 48 39 10 12

– 54 14 11 38 11

– – 38 39 25 6

– – – 12 14 33

– – – – 13 29

– – – – – 9

Note. AIC: Akaike Information Criterion; BIC: Bayesian Information Criterion; ssaBIC: sample-size adjusted Bayesian Information Criterion; LMR-ALRT: Lo–Mendell–Rubin adjusted loglikelihood ratio test.

Response Probability

A. Peacock, R. Bruno / Addictive Behaviors 45 (2015) 252–258

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

255

3.4. Correlates of Consumer Group Membership

Low Risk-Taking Consumers (n=152) Disinhibited Intake Consumers (n=195) High Risk-Taking Consumers (n=56)

Fig. 1. Response probabilities according to group for the 3-class solution.

3.3. Latent class probabilities and class definitions Response probabilities for each class are shown in Fig. 1. Low risktaking consumers (38% of the sample) comprised people who had a low probability of engaging in any of the risk behaviours (probabilities of engagement in any examined risk behaviour all ≤ 0.19) in AmED drinking sessions. Disinhibited intake consumers (48%) comprised people who had a high probability of drinking more than they intended (0.85) and spending more than intended (0.75) in AmED drinking sessions; other risk behaviours (e.g., illicit drug use, no contraception during sex) were less likely to be reported by this group (all probabilities of engagement ≤0.36). High risk-taking consumers (14%) comprised people who engaged in a variety of risk-taking behaviours when consuming AmED. They had a high probability of drinking more than intended (0.93) and spending more than intended (N0.99), as well as smoking cigarettes (0.74), behaving in a way that resulted in them experiencing guilt (0.71) and humiliation (0.77), and being hurt or injured (0.63). These consumers also had a strong probability of using illicit drugs (0.53), having sex with a stranger (0.56) or failing to use contraception (0.53), verbally fighting with someone (0.58), and being physically hurt or injured (0.63).

3.4.1. Disinhibited intake consumers versus low risk-taking consumers Whilst there was no significant difference between disinhibited intake and low risk-taking consumers on the basis of age, education, or employment, the former group had significantly higher odds of being male (albeit still predominantly female: 56%) (Table 3). Disinhibited intake consumers had an almost four-fold increased likelihood of reporting frequent (fortnightly or greater) AmED consumption, as well as significantly greater alcohol and ED intake in AmED sessions. In both groups, median ED intake did not exceed the maximum daily recommended intake (i.e., no more than two standard 250 mL EDs; Food Standards Australia and New Zealand, 2009); in contrast, alcohol intake for both groups was higher than recommended to minimise risk of injury or harm (i.e., ≤ 4 standard alcoholic drinks; National Health and Medical Research Council, 2009). Disinhibited intake consumers also had a significantly higher median trait impulsivity score than low risk-taking consumers. 3.4.2. High risk-taking consumers versus low risk-taking consumers Similarly, there was no significant difference between high risktaking and disinhibited intake consumers on the basis of age, education, or employment; the former group had two-fold increased likelihood of being male (nearly half the group) (Table 3). Half of the high risk-taking consumers reported fortnightly or more frequent AmED use, leading to an eleven-fold increased odds of regular AmED use amongst this group relative to low risk-taking consumers. High risk-taking consumers had significantly greater alcohol and ED intake in AmED sessions, with median ED and alcohol intake exceeding recommended intake guidelines. As expected, this group also recorded significantly higher trait impulsivity scores than low risk-taking consumers. 3.4.3. Risk-taking in alcohol drinking sessions according to AmED risktaking typology Overall, the pattern of risk engagement was consistent for alcohol and AmED drinking sessions (Fig. 2). The odds of retrospectively selfreporting each risk-behaviour in alcohol drinking sessions were consistently greater for the disinhibited intake and high risk-taking consumers relative to the low risk-taking consumers (Table 4). For example, the disinhibited intake consumers had four-fold higher odds and the high risk-taking consumers had 28-fold increased odds of

Table 3 Demographic, alcohol and ED use, and trait impulsivity outcomes according to group membership (N = 403). Outcomea

(A) Low risk-taking consumers n = 152

(B) Disinhibited intake consumers n = 195

(C) High risk-taking consumers n = 56

B vs A (ref)b OR (95% CI)

Age (median, IQR)# % Male* % Unemployed* % Not commenced/completed university qualification % AmED frequent use* AmED ED standard drink intake (median, IQR)# AmED alcohol standard drink intake (median, IQR)# I7 Trait Impulsivity Subscale score (median, IQR)#

22 (20–24) 31 (24–38) 12 (8–18) 17 (12–24)

22 (20–25) 44 (37–51) 18 (13–24) 21 (16–28)

22 (20–24) 46 (33–59) 13 (6–24) 23 (14–36)

1.01 (0.95–1.06) 1.76 (1.12–2.75) 1.65 (0.89–3.05) 1.35 (0.78–2.34)

9 (1–14) 1 (1–2) 5 (2–7)

26 (76–87) 2 (1–4) 6 (4–10)

52 (39–64) 3 (2–4) 8 (4–12)

3.71 (1.93–7.14)*** 1.44 (1.21–1.70)*** 1.10 (1.04–1.15)***

11.32 (5.22–24.53)*** 1.66 (1.36–2.03)*** 1.15 (1.08–1.22)***

5 (3–8)

7 (5–12)

9 (6–12)

1.17 (1.11–1.24)***

1.22 (1.13–1.33)***

a

C vs A (ref)b OR (95% CI)

0.99 (0.92–1.08) 1.90 (1.01–3.59)* 1.06 (0.42–2.70) 1.51 (0.71–3.21)

Age was calculated in years; employment was classified as any engagement in full/part-time/casual work or paid study; AmED frequent use was defined as use on a fortnightly or more frequent basis in the past six months; the reference period for typical energy drink (ED; one standard drink = 250 mg beverage containing approximately 80 mg caffeine) intake in AmED sessions was the last six months; the reference period for typical alcohol (one standard drink = 10 g alcohol) intake in AmED sessions was the last six months; I7 Trait Impulsivity scores range from 0 to 19, with higher scores indicating greater trait impulsivity; * denotes binary categorical outcomes with percentage of endorsing participants and 95% confidence intervals in parentheses; # denotes continuous variables with the median and inter-quartile range in parentheses (all continuous variables demonstrated positive skew and/or kurtosis). b Univariate multinomial logistic regression results are presented here. An odds ratio of 1 indicates that the event is equally probable in each group, N1 indicates that the event is more likely to occur in the non-reference group relative to the reference group, and b1 indicates that the event is less likely to occur in the non-reference group; relative to the reference group. Mann–Whitney U tests were conducted for continuous variables: B vs A: age Z = −0.41, ED standard intake: Z = −3.76***, alcohol standard intake: Z = −4.42***, I7 score: Z = −4.74***; C vs A: age Z = 0.20, ED standard intake: Z = −5.23***, alcohol standard intake: Z = −4.62***, I7 score: Z = −4.62***. OR: odds ratio; 95% CI: 95% confidence interview; IQR: interquartile range; *p b .050; **p b .010; ***p b .001.

A. Peacock, R. Bruno / Addictive Behaviors 45 (2015) 252–258

Analysis of AmED-related risk-taking behaviour post-AmED consumption retrospectively self-reported by an Australian community sample indicated three qualitatively different classes of consumers. Low risk-taking AmED consumers (two-fifth of the sample) were unlikely to report any of the assessed risk-taking behaviours whilst under the influence of AmED. Disinhibited intake AmED consumers (half the sample) had a high probability of reporting excess alcohol intake and excess financial outlay whilst intoxicated; low level endorsement of other risk-taking behaviours (e.g., sexual risk-taking, verbal aggression, physical injury) was evident. High risk-taking AmED consumers (over one-tenth of the sample) had a high likelihood of reporting all risktaking behaviours assessed: excess alcohol intake, financial risktaking, licit and illicit drug use, sexual risk-taking, behaving in a way resulting in psychological distress, and causing injury or harm to self. The profiles of each group were quite distinct: disinhibited intake consumers and high risk-taking consumers were more likely to report: i) being male, ii) hazardous alcohol use (higher AUDIT score), iii) more frequent AmED use, iv) higher ED intake in AmED sessions, v) higher alcohol intake in AmED sessions, and vii) higher trait impulsivity (higher I7 score), relative to low risk-taking consumers. AmED behavioural patterns were reflected in analysis of risk-taking behaviour in alcohol drinking sessions. Disinhibited intake consumers reported higher odds of each risk behaviour relative to low risk-taking consumers. Notably, greater odds of risk-taking behaviour were observed for high-risk consumers relative to low risk-taking consumers, with 28-fold increased odds of being physically hurt or injured. Overall, these results align with previous research showing heterogeneity amongst the consumer group in the behavioural consequences of AmED use (Mallett et al., 2014). A growing body of research shows that AmED consumers report greater engagement in risk-taking compared to alcohol consumers (e.g., Brache & Stockwell, 2011; O'Brien, McCoy, Rhodes, Wagoner, & Wolfson, 2008), and this association between AmED use and behavioural risk-taking has been reinforced through popular media (e.g., Rechter, 2012). However, the present study showed that the majority of consumers can be defined by: i) low engagement in risk-taking behaviour during AmED sessions, or ii) drinking too much and spending too much. The finding that half the sample fell into the latter category is not surprising given that these two behaviours are most often endorsed amongst alcohol consumers (Lubman et al., 2013). However, a significant minority of AmED consumers reported engaging in a spectrum of hazardous and potentially harmful behaviours; these behaviours were evident in AmED and alcohol drinking sessions. Engagement in most assessed risk-taking behaviours by the high risk group is not surprising considering these consumers report more frequent AmED and alcohol use, and greater intake (although it is important to note that the frequency of these behaviours was not ascertained, as participants reported only whether the behaviour had occurred or not). However, these findings suggest that treating AmED consumers as a homogenous group could mask the hazardous behavioural profile evident post-consumption of alcohol or AmED for a significant few.

Low Risk-Taking Consumers Percentage of AmED Consumers

4. Discussion

From a research perspective, these results highlight the need to factor distinctive behavioural patterns across the consumer group into analyses when assessing the harms associated with this consumption trend. From a public health perspective, the current results indicate that policies and programmes to reduce potential behavioural harms associated with AmED use may need to consider the heterogeneous nature of the consumer group (Mallett et al., 2014). In particular, strategic awareness and intervention programmes aimed at reducing hazardous AmED use and behavioural harms may need to be targeted at this subgroup of high-risk consumers, rather than developed and delivered on a universal basis. Targeted interventions for high-risk alcohol consumers have demonstrated greater efficacy, lower costs, and greater compliance compared to general health promotion strategies (Toumbourou et al., 2007). Before planning targeted AmED harm minimisation endeavours, it should be noted that high risk-taking group membership was associated with higher odds of hazardous alcohol use, greater frequency of AmED sessions, excess ED intake in AmED sessions, greater alcohol intake in AmED sessions, and higher odds of risk-taking in alcohol drinking sessions. Consumers with higher trait impulsiveness similarly engage in risk-taking behaviour across AmED and alcohol drinking sessions, indicating that harm minimisation endeavours might be better focused on the individual, the frequency of alcohol use and intake, and their propensity towards risk-taking.

Alcohol Drinking Sessions AmED Drinking Sessions

60 40 20 0

Disinhibited Intake Consumers Percentage of AmED Consumers

being physically hurt or injured relative to the low risk-taking consumers. These two groups of consumers also reported more frequent alcohol use and greater typical intake on drinking occasions (excluding AmED sessions). Disinhibited intake consumers had significantly higher AUDIT scores than low risk-taking consumers (although it should be noted that the median AUDIT score for both groups was higher than 8, the recommended cut-off indicative of hazardous and harmful alcohol use; Babor et al., 2001). High risk-taking consumers had significantly higher AUDIT scores than low risk-taking consumers, with the median score exceeding 16, the cut-off indicative of a high level of alcohol problems and need for brief counselling and monitoring by a health professional (Babor et al., 2001).

Alcohol Drinking Sessions AmED Drinking Sessions

100 80 60 40 20 0

High Risk-Taking Consumers Alcohol Drinking Sessions Percentage of AmED Consumers

256

100

AmED Drinking Sessions

80 60 40 20 0

Fig. 2. Percentage of low risk-taking (top panel), disinhibited intake (middle panel), and high risk-taking (bottom panel) AmED consumers reporting risk-taking behaviours in AmED and alcohol drinking sessions in the preceding six months (N = 403).

A. Peacock, R. Bruno / Addictive Behaviors 45 (2015) 252–258

257

Table 4 Alcohol drinking session behaviour according to group membership (N = 403). Risk behavioura

N

(B) Disinhibited intake consumers n = 195

(C) High risk-taking consumers n = 56

B vs A (ref) OR (95% CI)b C vs A (ref) OR (95% CI)b

% Frequent alcohol use*,^ Typical alcohol intake (alcohol sessions) (median, IQR)#,^ AUDIT score (median, IQR)# % Smoked cigarettes* % Drank more alcohol than planned* % Used illegal drugs* % Had sex with someone recently met* % Did not use contraception* % Spent more money than planned* % Verbally fought* % Acted in a way that resulted in me experiencing guilt* % Acted in a way that resulted in me experiencing humiliation* % Physically hurt or injured*

400 91 (87–93) 86 (79–90) 395 6 (3–8) 4 (2–7)

93 (89–96) 6 (4–9)

95 (85–98) 8 (6–10)

2.34 (1.14–4.82)* 1.08 (1.02–1.14)**

375 401 397 397 393 392 398 397 400

9 (6–12) 33 (26–41) 53 (45–61) 16 (11–23) 18 (13–25) 12 (7–18) 55 (47–63) 17 (12–24) 26 (20–33)

12 (10–16) 46 (39–53) 88 (82–92) 29 (23–55) 32 (26–39) 28 (22–35) 83 (78–88) 37 (30–44) 57 (50–64)

19 (15–24) 78 (66–87) 86 (74–92) 67 (53–78) 73 (60–83) 65 (52–76) 95 (85–98) 60 (47–72) 82 (70–90)

1.18 (1.12–1.25)*** 1.73 (1.11–2.69)* 6.13 (3.59–10.45)*** 2.12 (1.24–3.63)** 2.12 (1.26–3.55)** 2.94 (1.62–5.33)*** 4.06 (2.47–6.68)*** 2.78 (1.66–4.65)*** 3.81 (2.40–6.05)***

1.39 (1.29–1.50)*** 7.31 (3.55–15.08)*** 5.14 (2.28–11.62)*** 10.58 (5.18–21.62)*** 11.95 (5.79–24.68)*** 14.09 (6.63–29.92)*** 13.99 (4.18–46.80)*** 7.21 (3.64–14.31)*** 13.21 (6.08–28.67)***

399 45 (40–50) 27 (20–34)

49 (42–56)

82 (70–90)

2.66 (1.68–4.21)***

12.77 (5.89–27.67)***

396 27 (23–32)

27 (21–34)

75 (62–84)

3.58 (1.90–6.76)***

28.24 (12.45–64.06)***

Total sample

12 (8–16) 45 (41–50) 75 (70–78) 29 (25–34) 33 (28–37) 27 (23–31) 75 (70–78) 32 (28–37) 49 (44–54)

(A) Low risk-taking consumers n = 152

9 (6–15)

2.93 (0.84–10.22) 1.17 (1.09–1.25)***

a

* denotes binary categorical outcomes with percentage of endorsing participants the behaviour in the last six months (or last month where indicated by ^) and 95% confidence intervals in parentheses; # denotes a continuous variable with the median and inter-quartile range in parentheses (all continuous variables demonstrated positive skew and/or kurtosis) Frequent use was defined as ‘fortnightly or more often’. b Univariate multinomial logistic regression results are presented here. An odds ratio of 1 indicates that the event is equally probable in each group, N1 indicates that the event is more likely to occur in the non-reference group relative to the reference group, and b1 indicates that the event is less likely to occur in the non-reference group; relative to the reference group. Mann–Whitney test was conducted for continuous variables: B vs A: typical alcohol intake: Z = −3.51***, AUDIT score: Z = −6.99***; C vs A: typical alcohol intake: Z = −5.21***, AUDIT score: Z = −8.36***. OR: odds ratio; 95% CI: 95% confidence interval; IQR: interquartile range *p b .050, ** p b .010, ***p b .001.

Regardless, the co-occurrence of hazardous alcohol and ED use with engagement in a broad array of risk-taking behaviours presents an alarming picture in regard to the future health and social problems for this group. Replication of these outcomes is necessary as this study comprised the first assessment of risk-taking behaviour profiles amongst this consumer group, with limitations inherent in the methodology and analytic approach. Data were self-reported and thus subject to potential bias (although a web-based survey was used to allow independent completion, and non-identifying information was collected to assure anonymity). The sample may not be fully representative of the community as participants were self-selected as part of a strategic recruitment strategy; this recruitment method allowed assessment of AmED consumption beyond the university student drinking culture. Furthermore, inclusion criteria restricted the sample to consumers who meet the legal drinking age for consuming or purchasing alcohol in Australia (18 years of age). Capturing prevalence data across all age groups in future research could clarify whether these class categorisations hold true across age groups. In regard to limitations stemming from the analytic approach, the current study was restricted to higher prevalence risk behaviours which were typically legal with potentially remote (e.g., failed to use contraception) or less severe (e.g., spent more money than planned) consequences. Lower prevalence behaviours which were typically illegal (e.g., driving a motor vehicle whilst intoxicated) or had severe immediate consequences (e.g., requiring emergency medical treatment) were not included due to insufficient statistical power for comparisons. Assessment of AmED consumer profiles with a larger sample is required so that the applicability of the model when immediate and/or severe consequence risk behaviours are included can be tested. AmED consumers are often viewed as the more hazardous, risky counterparts of alcohol consumers. However, this study adds to the growing body of literature indicating that AmED consumers may not be a homogenous group in regard to their risk-taking behaviour, with three types of consumers identified on the basis of behavioural patterns whilst under the influence of AmED. High likelihood of risk-taking behaviour in AmED sessions amongst a significant minority of AmED consumers, coupled with indications of alcohol misuse, higher

frequency AmED use, excess alcohol and ED intake, and greater alcohol-related risk-taking, emphasises the need for targeted policies and programmes to minimise future social and health problems. Role of funding sources There was no direct or indirect financial support for this research. Contributors AP and RB were responsible for the design of the study. AP was responsible for data collection, statistical analysis, interpretation of the data, as well as writing the first draft of the manuscript. All authors have contributed to and have approved the final manuscript. Conflict of interest AP and RB were provided placebo samples by Red Bull GmbH in a prior experimental study; no financial support was provided and this organisation had no involvement in design, interpretation, or reporting of the previous work. Red Bull GmbH has no involvement in the current manuscript. AP and RB are in receipt of an untied educational grant from Mundi Pharma for post-marketing surveillance of an oxycodone reformulation. The funders had no input in the design, interpretation or analysis of that study. The authors declare that there are no conflicts of interest for this research. Acknowledgements The authors wish to thank Associate Professor Frances H. Martin for her contribution to the design of the survey.

References Attila, S., & Cakir, B. (2011). Energy-drink consumption in college students and associated factors. Nutrition, 27(3), 316–322. Australian Medical Association (2010, December). AMA pushes for alcoholic energy drink ban. Available from: http://www.abc.net.au/news/2010-12-13/ama-pushes-foralcoholic-energy-drink-ban/2372020 Babor, T.F., Higgins-Biddle, J.C., Saunders, J.B., & Monteiro, M.G. (2001). The Alcohol Use Disorders Identification Test: Guidelines for use in primary care. World Health Organisation. Brache, K., & Stockwell, T. (2011). Drinking patterns and risk behaviors associated with combined alcohol and energy drink consumption in college drinkers. Addictive Behaviors, 36(12), 1133–1140. de Haan, L., de Haan, H.A., van der Palen, J., Olivier, B., & Verster, J.C. (2012). Effects of consuming alcohol mixed with energy drinks versus consuming alcohol only on overall alcohol consumption and negative alcohol-related consequences. International Journal of General Medicine, 5, 953–960.

258

A. Peacock, R. Bruno / Addictive Behaviors 45 (2015) 252–258

Department of Health and Ageing (2011, March). Ministerial Council on Drug Strategy Communique: 25 February 2011. Available from: http://www.health.gov.au/ internet/drugstrategy/publishing.nsf/Content/mcds-comm-feb11#har Eysenck, S.B.G., Pearson, P.R., Easting, G., & Allsopp, J.F. (1985). Age norms for impulsiveness, venturesomeness and empathy in adults. Personality and Individual Differences, 6(5), 613–619. Ferreira, S.E., de Mello, M.T., Pompeia, S., & de Souza-Formigoni, M.L. (2006). Effects of energy drink ingestion on alcohol intoxication. Alcoholism: Clinical and Experimental Research, 30(4), 598–605. Food Standards Australia and New Zealand (2009). Australia New Zealand Food Standards Code: Standard 2.6.4. Formulated caffeinated beverages. Department of Health and Ageing (Available from: http://www.comlaw.gov.au/Details/F2009C00814). Lo, Y.T., Mendell, N.R., & Rubin, D.B. (2001). Testing the number of components in a normal mixture. Biometrika, 88(3), 767–778. Lubman, D., Peacock, A., Droste, N., Pennay, A., Miller, P., Bruno, R., et al. (2013). Alcohol and energy drinks in NSW. Mallett, K.A., Marzell, M., Scaglione, N., Hultgren, B., & Turrisi, R. (2014). Are all alcohol and energy drink users the same? Examining individual variation in relation to alcohol mixed with energy drink use, risky drinking, and consequences. Psychology of Addictive Behaviors, 28(1), 97–104. National Health and Medical Research Council (2009). Australian guidelines to reduce health risks from drinking alcohol. Canberra: National Health and Medical Research Council. O'Brien, M.C., McCoy, T.P., Rhodes, S.D., Wagoner, A., & Wolfson, M. (2008). Caffeinated cocktails: Energy drink consumption, high-risk drinking, and alcohol-related consequences among college students. Academic Emergency Medicine, 15(5), 453–460.

Patrick, M.E., Evans-Polce, R.J., & Maggs, J.L. (2014). Use of alcohol mixed with energy drinks as a predictor of alcohol-related consequences two years later. Journal of Studies on Alcohol and Drugs, 75(5), 753–757. Peacock, A., Bruno, R., & Martin, F.H. (2012). The subjective physiological, psychological, and behavioral risk-taking consequences of alcohol and energy drink co-ingestion. Alcoholism: Clinical and Experimental Research, 36(11), 2008–2015. Peacock, A., Bruno, R., & Martin, F.H. (2013). Valid points, but the trends remain: A response to Rossheim, Suzuki, and Thombs. Alcoholism: Clinical and Experimental Research, 37(12), 2171–2174. Peacock, A., Pennay, A., Droste, N., Bruno, R., & Lubman, D. (2013). ‘High’ risk? A systematic review of the acute outcomes of mixing alcohol with energy drinks. Addiction, 109(10), 1612–1633. Pennay, A., & Lubman, D.I. (2012). More Australian research needed into alcohol and energy drinks. Drug and Alcohol Review, 31(7), 928–929. Ramo, D.E., Grov, C., Delucchi, L., Kelly, B.C., & Parsons, J.T. (2010). Typology of club drug use among young adults recruited using time–space sampling. Drug and Alcohol Dependence, 107(2–3), 119–127. Rechter, J. (2012). Call time on dangerous energy drinks. The Sydney morning herald. Toumbourou, J.W., Stockwell, T., Neighbors, C., Marlatt, G.A., Sturge, J., & Rehm, J. (2007). Interventions to reduce harm associated with adolescent substance use. Lancet, 369(9570), 1391–1401. United States Food and Drug Administration (2010, November). Serious concerns over alcoholic beverages with added caffeine. Available from: http://www.fda.gov/ ForConsumers/ConsumerUpdates/ucm233987.htm World Health Organisation (2010). Global strategy to reduce harmful use of alcohol. World Health Organisation (Available from: http://www.who.int/substance_abuse/ alcstratenglishfinal.pdf?ua=1).