Boon B., Risselada A., Huiberts A. et al.
Journal of Medical Internet Research: 2011, 13(2), e43.
Unable to obtain a copy by clicking title? Try asking the author for a reprint by adapting this prepared e-mail or by writing to Dr Boon at firstname.lastname@example.org. You could also try this alternative source.
Spending just ten minutes each on a drinking feedback and advice web site is leading over 2000 heavy drinking Dutch men a year to reduce to safer levels was the implication of this randomised trial from the Netherlands.
Summary Face-to-face interventions are limited by the shortage of suitable professionals, and heavy drinkers may be reluctant to discuss their drinking, meaning most remain untreated. In contrast, on-line interventions need require no therapist time and users can engage in them whenever they wish and in the privacy of their homes without fear of stigmatisation. This study aimed to examine the effect of DrinkTest, an on-line intervention targeting heavy drinking adults in the Netherlands by offering personalised feedback Personalised feedback is assumed to be more effective than general information because the information is perceived as more personal and therefore more relevant, meaning the recipient pays more attention to the key message. about each user's drinking pattern.
Such interventions have proven a feasible way to reach heavy drinkers and are generally well received. However, beyond student samples, results are not yet conclusive. An analysis synthesising the results of on-line alcohol and tobacco interventions included only three studies of drinking among general adult populations. Another research synthesis focused on drinking reported an overall medium effect size A standard way of expressing the magnitude of a difference (eg, between outcomes in control and intervention groups) applicable to most quantitative data. Enables different measures taken in different studies to be compared or (in meta-analyses) combined. Based on expressing the difference in the average outcomes between control and experimental groups as a proportion of how much the outcome varies across both groups. The most common statistic used to quantify this difference is called Cohen's d. Conventionally this is considered to indicate a small effect when no greater than 0.2, a medium effect when around 0.5, and a large effect when at least 0.8. for nine randomised controlled trials, including the present study, but most of the on-line interventions were fairly time consuming, ranging from one 90-minute session to a 10-week programme, and some required therapist involvement.
Developed by the Netherlands Institute for Health Promotion and Disease Prevention, DrinkTest is delivered in a single 10-minute session with no therapist involvement. It asks the user about their usual weekly alcohol intake and their intentions to change their drinking. Based on this it offers feedback on their intake and associated health risks compared to other Dutch residents of the same age and sex. Users who score as heavy drinkers (by design, all did in the current study) are further quizzed about their drinking occasions and patterns, how confident they feel in their ability to change, and their attitudes and intentions with regard to reducing alcohol intake. Based on their answers, respondents receive personalised feedback on how to reduce alcohol intake in their specific situation, which in the study they could print and take home.
An initial study found DrinkTest worked for women but not men, leading to the revision tailored to men tested by the featured study. Participants were recruited via an on-line screening questionnaire sent to around 9000 men on two survey panels and through national newspaper adverts. In all 887 respondents whose drinking exceeded Dutch guidelines Men who consumed more than 200gm of alcohol per week and/or more than 50gm on a single occasion on at least one day per week. but had not recently been treated for this were invited to take part in the study. Of these 450 agreed and were randomly allocated to the DrinkTest intervention or (the control A group of people, households, organisations, communities or other units who do not participate in the intervention(s) being evaluated. Instead, they receive no intervention or none relevant to the outcomes being assessed, carry on as usual, or receive an alternative intervention (for the latter the term comparison group may be preferable). Outcome measures taken from the controls form the benchmark against which changes in the intervention group(s) are compared to determine whether the intervention had an impact and whether this is statistically significant. Comparability between control and intervention groups is essential. Normally this is best achieved by randomly allocating research participants to the different groups. Alternatives include sequentially selecting participants for one then the other group(s), or deliberately selecting similar set of participants for each group. group) to read and if they wished take home a standard brochure on the biological effects of alcohol and healthy and unhealthy drinking patterns. Rather than the study's true objective, participants were told they were in a study of healthy lifestyle education and had happened to be allocated to the alcohol topic. Both the on-line intervention and reading the brochure were completed at research offices rather than at home.
On average the sample was 40 years of age and drank about 315gm alcohol a week or 39 UK units, and most were employed and well educated.
Of primary interest was the proportion of men who cut their drinking to below Dutch guidelines for low-risk drinking. One month after working through DrinkTest, 42% had compared to 31% who had read the brochure. The proportions equated to one person becoming a 'safe' drinker for every nine allocated to DrinkTest. This difference remained statistically significant when another method was used to estimate the drinking of participants who did not complete the follow-ups and when the analysis was confined only to respondents.
Six months after the interventions, still more had become 'safe' drinkers, but the gap had shrunk to 46% v. 37% and now narrowly missed remaining statistically significant chart (this recalculates the figures to the proportions still drinking excessively). Assuming a real impact at six months, the difference equated to one person becoming a 'safe' drinker for every 12 allocated to DrinkTest.
Impacts of the interventions did not significantly differ for people of different ages, educational attainment, or average weekly drinking.
At least in the short-term, computer-based personalised feedback led more male heavy drinkers to cut their intake to safer levels than a standard alcohol information brochure. These findings derive from a study with a high follow-up rate and were robust under alternative analytic methods. The initial impact was comparable to that found after a more intensive on-line, four-step cognitive behavioural intervention also directed at the Dutch general adult population.
The intervention web site attracts about 90,000 men a year of whom 70% score as heavy drinkers. Four of ten complete the test and presumably receive tailored feedback. To estimate the potential impact, it is assumed that this usage is maintained and that the impact at six months in the study is a valid indicator of the impact in the home setting. Then each year 2117 men would cut their alcohol intake for at least six months as a consequence of spending 10 minutes on DrinkTest. In turn this suggests that offering personalised feedback through highly accessible web sites can efficiently and economically contribute to generating health gains at the population level. This suggestion is reinforced by a study modelling the potential impact in the Netherlands of alcohol eHealth interventions such as DrinkTest. Moreover, DrinkTest's impact was consistent across different types of heavy drinkers, again a finding which replicates that for another Dutch on-line alcohol intervention.
Of note is that even in the control condition many participants curbed their drinking. High levels of motivation among those who agreed to participate seems an unlikely explanation as they thought they were engaging in a lifestyle education study, not one specifically to do with drinking. Perhaps repeated questions about drinking at different time points had an impact. If so, the on-line intervention still exerted an extra impact.
While effectiveness has been demonstrated in 'laboratory' conditions, it remains to be seen whether these results hold up when the intervention is (as normally it would be) engaged in at home, where web site users might not complete it or pay as much attention to its advice. A hopeful sign is that a synthesis of similar studies did not find on-line alcohol interventions any less effective when undertaken at home as opposed to other locations. Moreover, any such effect could be expected to also increase the impact of reading the brochure.
commentary A rigorous analysis with little loss to follow-up gives confidence that the outcomes represent a valid extra impact on the participants recruited to the study of the on-line intervention relative to reading a brochure, in the context of a study which virtually ensured full attention to both. The main questions ( below) are over whether this impact would be replicated among the general run of heavy drinkers accessing the service in the normal way at home.
As acknowledged by the researchers, the home environment differs from the soundproofed-room research setting where participants sat with no distractions and only one task before them – to complete the on-line intervention or read the brochure, paying attention in expectation that their views would be sought as part of the study they had committed to. It is reasonable, as the authors point out, to suggest that if this magnified the impact of the interventions, it would have done so for both, yet still the on-line process led to greater drinking reductions. But it is also the case that if the on-line process had an advantage, this would itself probably have been magnified. Already in the study the advantage is quite small and possibly fleeting. In normal circumstances, it might be still less convincing.
A second concern is that a direct approach to 9000 people plus national newspaper ads netted just 413 who completed the one-month follow-up. Moreover, presumably they joined the study not because they were concerned or curious about their drinking – why people might visit the DrinkTest site or read a brochure – but because they wanted to help with a lifestyle education study. Whether how they reacted to the interventions represents how the general heavy-drinking adult Dutch population would react is an open question. It is, of course, entirely open in respect of women. They reacted well to the previous version of the on-line intervention, but it cannot be presumed they will still react well to the new male-adjusted version; if men reacted differently to the two versions, so too may women.
None of this is to seriously cast doubt on the validity of the impacts on the people who did participate in the study, or to deny the probability that others interested enough to access the intervention would respond similarly. However, it could be that rather than a resource accessed widely enough to have an impact on public health across a country, internet-based alcohol applications become one more niche option attracting and/or having a beneficial impact on a rather different population to conventional care.
Some of the limitations of the featured study broadly applied also to the evaluation of a Canadian version of a very similar 10-minute web-based intervention. The main exception was that this did test the intervention in the conditions in which it would normally be used: without face-to-face guidance and in the user's own home, or wherever else they chose. This too found initially promising results, but by 12 months later drinkers allocated to the intervention were drinking the same amount as those who had not been, and the significant extra drinking reductions seen earlier had disappeared. As well as confirming the fleeting nature of these reductions, that study offered an alternative explanation for the early findings: that intervention participants, having been alerted and sensitised to their excessive drinking, then under-reported it, creating the illusion that they had but back more than the control group.
The featured intervention was among those whose impacts were simulated for the Netherlands, the results of which suggested that national health would improve and/or intervention costs be reduced if on-line brief interventions and therapy were added to or replaced conventional alcohol-related health care. The other interventions were:
• DrinkingLess, an on-line four-step cognitive behavioural intervention involving exploring one's alcohol use, setting goals, changing behaviour, and maintenance of behaviour change;
• OnlineTreatment, an on-line therapist-led treatment for problem drinking; communication between participant and therapist is conducted over the internet in seven chat sessions of 45 minutes each covering setting goals, self-control techniques, monitoring, recognising relapse-precipitating situations, and relapse prevention techniques.
Since these three eHealth interventions increase in intensity, it was suggested that they could be used in a stepped-care framework, starting with the least intensive intervention, the DrinkTest, and if needed moving up to the more intensive levels of DrinkingLess and OnlineTreatment.
See other Findings analyses for a review of computer-delivered self-help interventions for drinking and smoking and a review focused on drinking. Both analyses include further commentary on the role of computer delivery and on UK findings.
Thanks for their comments on this entry in draft to John Cunningham of the Centre for Addiction and Mental Health in Toronto in Canada. Commentators bear no responsibility for the text including the interpretations and any remaining errors.
Last revised 05 May 2012
REVIEW 2012 Computer based alcohol interventions