Introduction

The UK Government recommends drinking should be kept within the lower risk category of 3–4 units for men and 2–3 units for women per day (UK unit = 8 g of alcohol) [1]. Excessive drinking, or binge drinking, is defined as consuming double this amount (≥8 units for men, ≥6 units for women) during one drinking occasion [2]. However, drinking over such guidelines is common, especially within college student populations. This issue is relevant to countries in several world regions (e.g. in Australasia, Europe, South American and North America [3]); in the US more than 80% of students drink alcohol and almost half report binge drinking in the last two weeks [4] (US binge definition: ≥5 drinks for men, ≥4 for women. Unit = 12 g alcohol [5]) and high levels of alcohol-related risk and harm (61%) are found in students within the UK [6]. With a constant increase in student numbers [7], college students represent a unique and high-risk group for alcohol-related harm and successful drink reduction strategies for this population would offer significant benefit for the individual and the wider society.

Policies regarding the dangerous use of alcohol often focus on harm-reduction methods. Current UK policy includes informing the public on alcohol guidelines so that individuals can make informed choices about healthier and responsible drinking, ultimately yielding a change in the broader drinking culture [8]. Information provision strategies have become widespread, but show disappointing results [9]. A review of mass media campaigns aimed at behavioural change concluded that in regards to alcohol consumption, they have had little success [10].

There are several possible explanations for the lack of effectiveness of information strategies. Firstly, people may not engage with or remember the information [11]. Jouriles and colleagues [12] found that recalling and writing down alcohol-related information led to a greater reduction in consumption after 2 weeks. Therefore increasing engagement with the information may mean it is remembered better and so is able to affect behaviour. Secondly, general harm-reduction strategies rarely fit the population as a whole, therefore policies may be more effective if personalised to the individual [13]. Brief personalised interventions (BPI) deliver short structured “brief advice” to encourage risky drinkers to lower their drinking levels [14], with content based on personalised information to increase an individuals’ awareness of their own behaviour [15]. Risky drinkers can be defined as ‘increasing risk’ drinkers (> 21 units for males and >14 units for females) or ‘higher risk’ drinkers (> 50 units of alcohol per week for males and > 35 units of alcohol per week for females) [16]. A web-based study involving Swedish students found that drinking assessment with personalised feedback resulted in fewer risky drinkers at 3 month follow-up compared with an assessment only and control group [17]. Similarly, a meta-analysis concluded that interventions containing personalised feedback are more effective than ‘assessment only’ control conditions in students [18]. However, compared to active controls (defined as brief-alternative alcohol-related interventions, general health interventions or interventions that provide only alcohol education) BPIs produced no additional benefit.

Research into BPIs in student populations has identified a number of elements which may help increase effectiveness: identifying consequences that are personally relevant; highlighting the practical costs of alcohol consumption; outlining risk reduction strategies; and including normative comparison [19]. All of these elements were contained in the BPI evaluated by the Screening and Intervention Programme for Sensible drinking (SIPS) trial [20]. This BPI was designed to provide practitioners with a structure to deliver brief advice to hazardous and harmful drinkers with an aim to capture risky drinking at an early stage, and provide advice or counselling to help reduce consumption [14]. SIPS assessed BPI effectiveness within a primary care setting but failed to find a difference between the BPI and control groups [20], perhaps indicating that different populations require different BPI elements.

Given that these BPI elements are not present in active controls, it is not clear why a difference in the effectiveness of these two approaches has not been found. It is possible that the personalised nature of the BPI elements encourages engagement with the intervention, and that it is this engagement which is key. The current study aimed to compare a BPI with an active control designed to encourage engagement (in the form of a quiz). Importantly, if both interventions are effective, this highlights that more cost effective harm reduction interventions (e.g., general information provision) may be effective in students, as long as engagement is successful.

One additional mechanism suggested to underlie drinking behaviour [21] is readiness to change (RTC), and some studies have found an increase in RTC scores following an intervention [22]. Therefore a further aim of BPIs in students may be to increase RTC [23]. In summary, the current study assessed the impact of a BPI, which includes the active elements previously identified as important in student populations, on alcohol consumption, frequency of binge drinking and readiness to change, compared with an active control (information-only) group in UK students. It was hypothesised that, compared to the active control, the BPI would result in (i) reduced fortnightly alcohol consumption and binge frequency and (ii) increased readiness to change.

Method

Results

Distribution of data was analysed and skewed variables were transformed to allow for parametric testing. There was one dropout in the control condition, who did not differ on baseline characteristics.

Participant Characteristics (see Table 1)

MANOVA investigated whether participants across the two groups differed across independent measures (age, readiness to change ruler, AUDIT, weekly consumption, weekly binge). There were no significant differences at baseline between the groups (ps>0.05) and the groups did not differ by gender χ²(1, N = 103) = 0.79, p = 0.38. The sample was made up of 88.35% risky drinkers, identified by an AUDIT score of 8 or above [27]. Groups did not differ in percentage of risky drinkers χ²(1, N = 103) = 0.42, p = 0.52.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Means (±SD) for participant characteristics by condition (N = 103).

https://doi.org/10.1371/journal.pone.0131229.t001

Changes in alcohol consumption

Changes in weekly alcohol unit consumption and binge frequency were investigated using mixed design 2x2 ANOVAs, with within-subject factors of time (2: baseline, follow-up) and between-subject factors of condition (2: BPI, active control).

The analysis of weekly unit consumption (Fig 1) revealed a significant main effect of time, F(1, 101) = 18.52, p = 0.001, ηp² = 0.16, with consumption decreasing significantly in both the BPI group, t(1, 51) = 3.24, p = 0.002 and the active control group t(1, 50) = 2.85, p = 0.006. The main effect of group, F(1, 101) = 0.48, p = 0.49 and the group x time interaction, F(1, 101) = 0.03, p = 0.86, were not statistically significant.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Weekly unit consumption at baseline and follow-up by condition.

https://doi.org/10.1371/journal.pone.0131229.g001

The analysis of binge frequency revealed an identical pattern (Fig 2). There was a significant main effect of time, F(1, 101) = 11.50, P = 0.001, ηp² = 0.10; binge frequency decreased in the BPI group, t(1, 51) = -2.41, p = 0.02 and the control group, t(1, 50) = 2.75, p = 0.01. The main effect of group, F(1, 101) = 0.38, p = 0.53 and the group x time interaction, F (1, 101) = 0.23, p = 0.63 were not statistically significant.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Weekly binge at baseline and follow-up by condition.

https://doi.org/10.1371/journal.pone.0131229.g002

Readiness to change (Fig 3)

Change in RTC was investigated using a mixed design 2x2 ANOVA, with within-subject factors of time (2: baseline, follow-up) and between-subject factors of condition (2: BPI, active control).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. RTC ruler at baseline and follow-up by condition.

https://doi.org/10.1371/journal.pone.0131229.g003

The analysis of RTC revealed a significant main effect of time, F (1, 101) = 5.7, p = 0.02, ηp² = 0.05, although post hoc t-tests indicated that the RTC increased between baseline and follow-up in the active control group, t(1, 51) = 2.05, p = 0.05, but not the BPI group, t(1, 51) = -0.97, p = 0.34. However, the main effect of group, F(1, 101) = 1.38, p = 0.24, and the group x time interaction, F(1, 101) = 1.06, p = 0.30, were not statistically significant.

Discussion

The present study compared the effect of a BPI with an active control on weekly alcohol consumption, binge frequency and readiness to change, in UK students. It was found that both groups significantly decreased their alcohol consumption and weekly binges to a similar extent.

The finding that both groups decreased their alcohol consumption significantly and by a similar amount suggests that the BPI offers no additional benefit to the active control [18]. This fits with previous research which has failed to find an effect of BPIs, compared with control, in reducing drinking within primary care settings [14]. It is possible that the SIPS’ BPI and control conditions may have contained similar active factors of behaviour change [35]; therefore the current findings may be due to specific components of the active control which included screening of quantity, frequency and consequences of the participants’ drinking [36]. Effects from assessment-only control groups is not a new phenomenon; small reductions on AUDIT scores and weekly alcohol consumption have been identified in control groups [37,38] and studies looking specifically at the effect of screening compared to non-screening have found significant effects on measures of hazardous drinking [39–41].

Most general information harm-reduction strategies simply provide individuals with various facts about alcohol and its harms. However, an important aspect of our study was that we encouraged engagement with the information-only component of the interventions by asking the participants to use the information to complete a quiz. Although research has suggested that information-only strategies are often ineffective [10], it is possible that they can be efficacious if engagement is maximised [12]. Information-only strategies are cheaper than BPIs [42] and so if we are to continue investing in information-only strategies, future research should identify ways to maximise engagement with the information. Our recall level at follow-up for both groups was approximately 50% which is significantly higher than that found by others (e.g, 10%) [12]. This suggests that quizzes may be one effective engagement strategy.

Another potential explanation for behaviour change in BPI and active control conditions is that both act upon the self-regulation of behaviour [38]. The process of reporting one’s own behaviour may result in reflecting on that behaviour which can trigger change. A possible mechanism is that by assessing our drinking behaviour, we become aware of risky practices, which initiates self-monitoring. Monitoring can allow us to recognise inconsistencies between current behaviour and a personal standard, which may impel us to change [43].

In terms of RTC, we hypothesised that the intervention would increase scores. However, there was no increase in the BPI group and there was no difference in RTC scores across groups. There have been mixed results with regards to interventions, alcohol use and motivation amongst students [44] with little success in regards to increases in RTC [45,46]. Evidence suggests that there is a non-linear relationship between alcohol use and RTC, as high levels of drinking are shown in those who show moderate levels of RTC, and low levels of drinking in those with high or low RTC [47]. A high variability on the RTC ruler both within and between individuals has also been found in previous research, with fluctuating alcohol consumption rates and scores on the ruler [48]. Taken together, research suggests RTC is a phenomenon difficult to measure and its relationship with alcohol is complex. It has been suggested an absence of consistent findings in this population of college students indicates the construct is of less importance for non-treatment seeking individuals [49].

There are a few points that need to be highlighted in terms of limitations. Firstly, most BPI research implements a screening method to identify risky drinkers before the intervention is administered [18], whereas the current study employed a universal approach. It could be argued that non-risky drinkers will show low RTC at baseline and therefore any effect of the intervention on RTC will be lost. However, over 88% of our sample were classed as risky drinkers so it is unlikely that RTC effects were missed. Additionally, one study found that abstinent students receiving personalised feedback were twice more likely to remain abstinent after one year than those in the control, indicating that in abstainers, feedback can function as a preventative measure [50]. Therefore, identifying which intervention components reduce risky drinking but also maintain non-risky drinking is important for future research. Although we note this was not an aim of this study and would require a longer follow-up period.

Secondly, the follow-up period of two weeks is a moderately short period to examine changes in drinking behaviour; therefore, firm conclusions regarding long-term effects cannot be drawn. Although many BPIs can be effective in the short term, the duration of this outcome is difficult to determine [19] with many studies demonstrating that drinking reduction effects diminish over time [51,52]. However, a recent meta-analysis of alcohol interventions for college students [18] found that effect size magnitude and intervention efficacy measures did not differ by assessment interval, apart from the frequency of heavy drinking, where it was found that when post-intervention assessments occurred one month or later they were more successful. Future research should include extended follow-up periods to determine how long any beneficial effects last and whether this differs across intervention type. If reductions do diminish over time, then future intervention development could involve ‘refresher’ interventions at regular intervals (e.g. via the internet or smartphone ‘apps’).

Importantly, as found in the majority of student studies [18], the current study demonstrates that in a UK student population a BPI offers no additional benefit to an active control at reducing alcohol consumption over a short time period. Given the risky drinking behaviour of this population, effective, evidence-based polices on alcohol harm-reduction strategies are needed. The active control, consisting of a drinking assessment and engagement with alcohol-related information, was as effective as a BPI in reducing alcohol-related behaviour. Active controls are arguably more cost effective than traditional BPIs which require more time and effort. If information-only interventions are suitable in this population, the next step will be to identify the best way to engage students in these strategies.

Supporting Information