Introduction

Alcohol is the most often used psychoactive substance by young adults and alcohol use is one of the leading modifiable morbidity and mortality risk factors among young adults [1,2]. The burden of alcohol among young individuals is mostly explained by a binge drinking pattern (the consumption of large quantities of alcohol over short periods of time) which is highly prevalent and is known to be associated with accidents, violence (hetero and self inflicted), sexual assaults and other negative consequences [2–4]. Therefore, from a public health point of view, preventing unhealthy alcohol use among young men in the general population would be valuable.

Several public health interventions have shown potential effects. Substance use prevention programs have been tested in schools with some evidence of efficacy on alcohol use initiation [5]. Structural measures are key for primary prevention, such as limited availability, price and tax increase, [6] but could be complemented with individualized interventions. There are however major challenges related to unhealthy alcohol use prevention, such as competing economical interests in selling alcohol containing products, the willingness to decrease all types of taxes in several jurisdictions, limited interest in developing or financing preventive interventions.

Large scale individualized intervention have been made possible with new technologies. Electronic and internet based interventions represent a potentially effective option, since they allow reaching broad populations by overcoming some of the logistic constraints of face to face interventions (geographical limitations, availability and training of counselors, delivery 24 hours a day at the subjects’ convenience). It is also a relatively inexpensive procedure [7]. Internet and computer-based electronic secondary interventions has been increasing over the past years. The same delivery method can be used for primary prevention interventions. Such interventions have already been tested with some success to prevent cannabis use [8]: a computer delivered intervention in primary care was shown to be associated with a significantly lower rate of cannabis use initiation 12 months later.

If there is growing evidence for the efficacy of internet secondary prevention brief interventions for unhealthy alcohol use [7,9–12], evidence is lacking for primary prevention interventions. Since internet based brief interventions can decrease alcohol use among people with unhealthy alcohol use, one could hypothesize that the approach might be suitable for preventing the development of unhealthy use, especially binge drinking. Nevertheless, one could also hypothesize that giving information on drinking to people who are not reporting unhealthy use (personalized feedback on the absence of consequences of use for example), or comparing the drinking of people who are drinking less to the more elevated drinking of people their age and sex (normative feedback for people reporting low use of alcohol or abstinence) might not be appropriate and might lead to increases in drinking. Indeed, an iatrogenic effect of primary prevention interventions has been questioned [13]. The question is of importance since people who are drinking less are likely to be exposed to internet based interventions when those are made available to large groups of people. Even though other studies have reported the absence of iatrogenic effect of preventive interventions among people who are drinking less, additional evidence is needed [14–16].

Therefore we conducted a randomized controlled trial (RCT) examining the efficacy of an internet based prevention intervention to prevent unhealthy alcohol use (i.e. primary prevention) compared to a control (no intervention) condition among young men who did not report unhealthy alcohol use. The goal of the intervention was to prevent an increase in alcohol use, prevent the development of monthly binge drinking or the development of alcohol use consequences.

We hypothesized that, at follow up, participants receiving the intervention would report lower alcohol use, would be less likely to report monthly binge drinking and less likely to report consequences of drinking compared to controls at 1 and 6 months.

Methods

Results

Among the CSURF participants, 4365 were invited to participate in the internet study. Of those, 724 refused, 2008 never answered the invitation (after two reminders), and 1633 agreed to participate. Of those, 737 reported unhealthy alcohol use (and were included in another study), leaving a sample of 896 for the primary prevention study reported here. Participants were randomized to the intervention (n = 451) or control group (n = 445). Of the 896 participants included, 844 and 835 completed the first and second follow up, respectively (94.2% and 93.2% follow up rate). The study flow chart is presented in Fig 1.

Download:

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

Fig 1. Study participant flow and follow-up rates.

https://doi.org/10.1371/journal.pone.0144146.g001

Baseline characteristics of participants are presented in Table 1. In accordance with the inclusion criteria, no participant had unhealthy alcohol use. Therefore, none reported binge drinking. The reported mean (SD) number of drinks/week was 2.3 (2.2), the mean (SD) AUDIT score was 3.5 (2.1) and participants reported a mean (SD) of 1.0 (1.1) alcohol related consequence over the past 12 months. No differences were observed on baseline variables between participants in the intervention and control groups (all p>0.2). No differences on baseline values were observed between those with and without follow up data (all p>0.2).

Download:

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

Table 1. Baseline characteristics of participants.

https://doi.org/10.1371/journal.pone.0144146.t001

Number of drinks per week

At 1 month, participants in the intervention group reported a mean (SD) of 2.3 (2.6) drinks per week compared to 2.8 (3.7) in the controls. At 6 months the reported consumption was 2.5 (3.0) and 2.7 (3.9) in the intervention and control groups, respectively. The results of the regression analysis are presented in Table 2. There was a borderline intervention by time interaction at 1 month, with participants in the intervention group reporting less drinking at 1 month (IRR = 0.87 (0.76; 1.00), p = 0.05).

Download:

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

Table 2. Regression model for number of drinks per week, baseline to 6 months (n = 896).

https://doi.org/10.1371/journal.pone.0144146.t002

Discussion

There is a lack of knowledge on the potential effects of primary prevention intervention for unhealthy alcohol use. Our study showed small but promising effects. We observed a statistically significant protective effect of the intervention on the number of drinks per week at 1 month but not at 6 months. We did not observe a statistically significant intervention effect on the binge drinking prevalence, nevertheless, we consider that a 6% difference in binge drinking prevalence in favor of the intervention group is notable, even though the effect did not reach a statistically significant level of 0.05. In addition, a protective effect was found on the occurrence of alcohol related consequences. This finding is at odds with electronic screening and brief intervention studies conducted among individuals with unhealthy alcohol use showing effects on drinking but not on alcohol related harm. Since the occurrence of alcohol related consequences was a secondary outcome, the impact of the intervention on consequences should be interpreted with caution.

An additional notable result is the absence of iatrogenic effects of the intervention. When internet-based interventions are made available to large populations, it may be difficult to exclude “low-risk” drinkers. Additional evidence is needed to confirm that normative feedback does not lead to an increase in drinking among people who are drinking less [14]. Our study shows that, when delivered accompanied by clear encouragements not to increase their current drinking, normative feedback did not lead to an increase in drinking. Our results are consistent with the absence of a “boomerang” effect of normative feedback showed by Prince and colleagues in four light drinking student samples [24].

Even though the observed effects were small, our results indicate potential interest for internet-based primary prevention interventions. The observed limited impact implies that, in its current form, this type of intervention cannot be seen as an alternative to other and more effective preventive measures (such as restriction to alcohol, ban on advertisement, etc.). Given the potential negative consequences of binge drinking and given its prevalence in the young population, especially among males, we think this type of intervention is of interest.

Due to the selection of people without binge drinking at baseline, we did observe an increase in binge drinking prevalence in both groups, which is consistent with a regression to the mean effect. Another explanation is that participants were assessed at the age binge drinking becomes more prevalent. In this age group in Switzerland, the prevalence of monthly binge drinking is 46% [25].

Limitations include self-assessment: because of the study design, other measures than self-assessments were not feasible and social desirability bias is possible. Another limitation is that the study was conducted in a sample consisting of males only and is therefore not indicative of what would happen in a female sample. More research is needed among young women. In addition, when we determined the study sample size, we hypothesized a larger intervention effect compared to the observed effect (due to the lack of primary prevention studies in our population we had to extrapolate from studies conducted in other populations, including in people with unhealthy alcohol use). The study was adequately powered for the hypothesized effect (80% power, 1.5 drinks per week difference). Since the observed effect was much smaller, a larger sample would have been necessary to maintain a 80% power. The study sample allowed a 63% power to detect the observed difference of 0.5 drinks per week at 1 month.

Our study showed limited effects which is in line with other studies that included primary prevention elements for substance use [14–16]. More research is needed to investigate whether or not these interventions can be modified as to lead to stronger effects.

The strength of the present study is the use of a general population sample, coming from diverse geographical and linguistic regions over two of the main Swiss linguistic regions. In addition, Switzerland has the interesting feature of comprising regions with drinking patterns similar to Nordic European countries (German speaking part) and Southern European countries (French speaking part). As such, individuals with varying drinking cultures and drinking environment were included, which is what would happen if such intervention is made available to the general population.

Supporting Information

Acknowledgments

N Sanchez: project manager; J. Stierli, S. Huber, and M. Dubi: research assistants; S. Szabo, J. Szabo, M. Grand d’Hauteville, and D. Gartner: web-design & development; M. Rège Walther: intervention development team; C. Eidenbenz: project manager C-SURF. The authors would also like to thank the participants for their time and dedication to the study.