Foxcroft D.R., Tsertsvadze A.
Cochrane Database of Systematic Reviews: 2011, 9, Art. No. CD009308.
Authoritative review finds that offering families of school-age children help to influence their drinking usually retards youth drinking – but typically effects are modest, may apply only to a minority of compliant and keen families, and it is questionable whether adding family work improves on well structured school drug education.
Summary The featured review conducted for the Cochrane collaboration analysed trials which randomly allocated participants to family-based programmes to prevent alcohol misuse in school-aged children aged up to 18 versus other types of interventions or no intervention. It was concerned with 'universal' Universal prevention interventions are appropriate when risk factors for development of a problem are not easy to identify, are diffuse in the population, and not easily targeted by an intervention. Another indication for universal programmes as opposed to selective programmes targeting high risk groups is when the 'prevention paradox' is a valid description of the distribution of problems – that is, more problems within a population arise from those at lower levels of risk than those at higher levels of risk. Several studies indicate that the prevention paradox is indeed relevant to youth drinking. programmes – those aimed at large groups such as an entire age range, whether or not they are known to be specially prone to substance use or problems. In family settings, universal prevention typically entails developing parenting skills including providing support, nurturing, establishing clear boundaries or rules, and monitoring children's activities. In one important respect, family-based programmes differ from those based in schools: rather than directly intervening with the young people, they intervene via their parents and family.
A previous review also conducted for the Cochrane collaboration had included relevant studies published up to 2002. Searches were conducted to identify further studies up to 2010. No language restrictions were applied. Twelve trials were found (11 from the USA and one from the Netherlands) with 14,595 participants, all published in peer-reviewed journals. Average ages of the children at the start of the trials ranged from 11 to 15.
Most trialled interventions aimed to raise the awareness of parents and children of issues such as the risks of substance use, and to promote social, behavioural and psychological changes in the children which would make substance use problems less likely. Among these were correcting overestimation of how 'normal' and accepted substance use is among their peers, boosting self-esteem, training in ways to resist other children's pressure to use drugs, and improving ability to solve problems and take decisions. Other features involved helping parents set rules and monitor and supervise their children, improving communication between parents and children, enhancing the quality of time spent together and attachment between family members, and reducing conflict.
To the extent that drinking and alcohol misuse are delayed, economic models calculate that some of the related long-term medical consequences of drinking too will be averted. This means that interventions which delay or curb drinking for several years are of more interest than those with short-term effects but no evidence of more persistent impacts. It was also intended to assess whether impacts differed for boys versus girls, children of different ages, those identified with different ethnic categories, Caucasian, black, Hispanic. or who at the start of the trial were drinking or not or drinking at different levels. In practice however, such analyses were not possible.
Nine of the 12 trials found statistically significant comparative reductions in drinking Measures included alcohol use initiation, mean composite index, frequency/quantity score of alcohol use, alcohol use or being drunk in past year, proportion of youth reporting lifetime alcohol use, alcohol use occasions, and initiation and frequency of drunkenness. among children allocated to family-based programmes as the sole type of intervention versus those allocated to no intervention or an alternative not involving family-based work. Follow-up periods ranged from two months to eight years, as did the duration of significant impacts.
Differences between the studies (in their interventions, subjects, and outcome measures) were such that it was not appropriate to pool their results. Instead these were described and salient features highlighted. Conclusions from this account are presented below.
The reviewed studies suggest that certain family-based prevention programmes can be modestly (but across a population, usefully according to economic models) effective and could be considered as policy and practice options. However, effect sizes and durations varied in ways which may depend on the content of the intervention and the context within which it is implemented, all the trials were conducted in western developed nations and all but one in United States, and methodological and reporting weaknesses make it difficult to absolutely rule out bias in the results of the individual trials and therefore in the findings of this review.
Nine of the 12 studies recorded statistically significant effects on drinking including over the longer as well as shorter term, and another found a positive effect which might have been significant had more families been included in the study. On the other hand, two studies with sufficient families to have found a positive effect in fact found none; one recorded apparently negative effects (which may have arisen by chance or due to methodological issues) and in the other, though ineffective on its own, when combined with a school-based intervention, a family-based intervention was more effective than comparison schools' standard curricula.
It could be that most of the positive studies and those recording no positive impacts reflect the underlying reality that family-based alcohol prevention programmes do not work, and that positive findings are due to chance variation around an overall zero impact. This is however unlikely given the preponderance of positive impacts and the sample sizes of the studies. More likely is that some (but not all) family-based psychosocial and developmental prevention interventions are effective in particular settings for reducing alcohol misuse among young people.
There is some evidence for the short to medium-term success of gender-specific interventions for daughters, typically involving their mothers. Two trials found impacts only among children already using substances at the start of the trials, findings perhaps best seen as requiring confirmation in trials designed for this purpose.
However, the worth of family-based prevention programmes does not rely solely on their impacts on drinking. Rather, they are intended to impact on a range of health and lifestyle behaviours among young people such as other substance use and antisocial behaviour.
commentary The cautious conclusions of this review – admitting the (though it was said, unlikely) possibility that the reviewed interventions are in fact ineffective – are warranted by what a British reviewer has described as the "dearth of methodologically highly sound research in this area" and the modesty of the observed impacts. An additional consideration is that typically such programmes have not been tested 'universally' in the normal sense of the word, but only on the sometimes few parents prepared to volunteer for the studies, engage in parenting interventions, and make themselves available to be followed up. The results are not necessarily a guide to what would happen if family-based programmes truly were made universal. For example, support for one of the most thoroughly researched family programmes comes mainly from a study whose findings derived from just over a third Though every family who participated and supplied the relevant data was included in the analysis regardless of whether they had attended the sessions. Also on all but one (parent education) of the variables measured (parent education, household income, target child gender, parent marital status, number of children, child conduct problems, and social-emotional distress), the families who participated in the study did not differ from those who did not. of the mainly white and rural families asked to participate in the study. A similar limitation applied to a later study of a substantially revised version among poor black families. Typically in Britain (see for example 1 2 3) and elsewhere in Europe, attendance for parent or family interventions is very low, especially among parents most in need of parenting support and with lenient attitudes to substance use.
Given the strong influence exerted by parental attitudes and behaviours on their children's drinking, it would however be a surprise if family programmes did not have some impact on children whose parents are willing to engage in the programmes and in the studies. Significant impacts were found by another review which, unlike the featured review, did amalgamate the results of relevant studies. In this case, published between 1995 and 2006. Across these it found that compared to alternative or no interventions, significantly fewer children allocated to family programmes started drinking during the follow-up periods and the average frequency of drinking too was reduced. But this review also had to warn that the results might not apply to families across the board but only those who fully engaged with the studies.
Given patchy outcomes and the great differences in the contexts and content of the interventions commented on by the featured review, there seems a clear need for a forensic examination of what might have led some programmes to work and others not, a procedure not attempted by the featured review on the basis that the published accounts did not give sufficient detail of what the interventions consisted of. Unfortunately this leaves practitioners in the dark about whether any of the approaches tried might work in their particular circumstances.
If a family programme does work it will do so largely by persuading and enabling parents who would not normally have done so to effectively control or influence their childrens' drinking. It seems likely that such interventions can only work well when they go with the grain of the society in which they are implemented, affording parental efforts legitimacy in their own and their childrens' eyes and offering tools the parents can use such as a strong probability of that their children will face disapproval from people they care about, impacts on school and work prospects, legal consequences for the child and perhaps too the parents, and high costs draining financial resources. These both give parents a reason to act and arguments to use other than simply, 'Don't do it – it is bad for you'.
The studies in the featured review generally pitted family interventions against no programme at all or a minimal one such as mailed advice leaflets. As might be expected among families apparently willing to engage in family interventions, actually offering them has more impact than perhaps disappointingly offering (virtually) nothing. Arguably the more meaningful question is whether with a limited prevention budget it makes sense to offer family programmes or to concentrate resources on other universal programmes, of which the most prominent is substance use education in schools. On this issue the evidence is thin and not on balance in favour of family or parenting programmes.
A companion review investigated programmes for the same purposes and populations as in the featured review, but which combined several components, typically school lessons and family/parenting interventions. It found no clear evidence that such multi-component interventions are more effective than single-component interventions. All the seven relevant studies added family/parental elements to direct intervention with the young people, the latter usually in the form of school lessons. In three there was no added impact. In another three there was, but two of these studies lacked a no-intervention group against which to assess whether any of the intervention combinations were more effective than usual practice. On examination, just one of these studies is at all persuasive of the added value of parental or family components. Details below.
The most convincing of the three positive studies was a Dutch trial which found that while each on their own did not improve on usual education, adding parenting components to a special classroom alcohol curriculum did substantially retard drinking among the 12–13-year-olds pupils. In this case the parenting element was built in to the schools' routine parent engagement programme, consisting of a brief presentation from an alcohol expert at the first parents' meeting at the start of each school year. It covered the adverse effects of youth drinking and the negative effects of permissive parental attitudes towards children's alcohol use, and was followed by collective or individual setting of rules on youth drinking by the parents.
In another study the extra affects on drinking of adding mailed cards to parents to reinforce brief advice from a nurse to their children was confined to the small minority of the average 13-year-old participants drinking at the start of the study and to one of the six alcohol use outcomes, results which given the number of outcomes tested for across drinkers and non-drinkers might have been a chance occurrence. Across the board, the greatest improvement in risk and protective factors related to drinking was actually seen in children allocated to the least intensive intervention focused on physical activity without any parental components.
The third study to find additional effects of parental components may not have trialled a universal intervention at all, because families were approached by local facilitators who used undocumented selection criteria and the families had to agree to participate. The fact that all the enlisted families (all black) engaged in the home-based family component – a video and role-play on monitoring children's activities and communicating about these between parent and child – suggests that considerable selection did take place. Without this component, after an eight-session group programme for the children (aged 13–16) the proportion drinking in the past six months increased at both six- and 12-month follow-ups to 31%. But among the families also offered the family component this increase was reversed, resulting at 12 months in just 22% having drunk. Whether this represents a true lasting impact of offering the family component seems questionable because at 12 months it was the only one of 13 outcomes which using suitable criteria and methods A less than one in 20 chance that the finding occurred by chance as the criterion for accepting there had been an impact; use of so-called two-tailed tests which do not effectively assume that if there had been a negative finding it must have been a meaningless fluke, a usually unwarranted assumption which doubles the probability of finding a significant effect. Another flaw in the analysis was the failure to adjust for the fact that the children were recruited at different sites and by different people. would have proved statistically significant, a finding which might have happened by chance.
Among the studies which found that parental/family components had no impact, the most surprising failure was the lack of persisting impact from adding probably the best established and most promising substance use prevention family programme – the Strengthening Families Programme – to a well structured and extensive school drug education curriculum. Despite earlier findings from the same study, in this US trial there was no real hint that working with the families improved substance use outcomes, though there may have been other benefits. Perhaps relevant is that only a quarter of the families allocated to these attended any of the family sessions, a programme which demanded the relatively heavy commitment of seven two-hour evening sessions plus four booster sessions.
Not included in the companion review was a seven-nation European trial which also found no extra benefits of adding parent workshops to school drug education; few parents attended, and an important element – role-play – was generally omitted.
Thanks for their comments on this entry in draft to Richard Velleman of the University of Bath in England. Commentators bear no responsibility for the text including the interpretations and any remaining errors.
Last revised 23 August 2012