The prevention of mental disorders in young people

There are nine RCTs of interventions to prevent depression in young people. Clarke et al4 used classroom teachers to deliver brief psychoeducation, and found no benefit. Petersen et al5 used psychologists and Spence et al6 used teachers in classroom settings to deliver extensive psychoeducation and problem-solving, and both found no benefit. Beardslee et al7 used clinician-assisted advice to a target group of children of depressed parents and found it not to be superior to brief information sessions.

The other five RCTs8-12 used variants of cognitive therapy or cognitive behavioural therapy (CBT) delivered by clinical psychologists. All these interventions led to improvement compared with the control groups, sometimes confined to symptoms and sometimes resulting in a reduced incidence of major depressive disorder. Clarke et al9 found the incidence of major depression in their intervention group of children of depressed parents was a third of that in the control group (9.3% v 28.8% over two years). Shochet et al,12 in a universal school-based intervention using a program designed by Clarke et al,8 found that the incidence of depression in the intervention group was halved over the 10-month follow-up.

The problem is not efficacy — the Clarke et al8 cognitive therapy program appears to work — but effectiveness in routine practice. There are too few clinical psychologists who are trained to deliver cognitive therapy and able to conduct either universal or targeted prevention programs in schools.

Two RCTs examined CBT programs for prevention in children at risk of developing an anxiety disorder.13-15 In the first of these, Dadds et al14 screened 1786 children aged 7–14 years, and identified 128 highly anxious children, who were then randomised either to the Barrett–Kendall16,17 10-session CBT intervention or to a waitlist control group. At the two-year follow-up, compared with the control group, the intervention group had half the risk of meeting criteria for an anxiety disorder (20% v 39%).

There are eight RCTs using a similar program in children and adolescents referred for early treatment of symptoms or disorder. They all show similar or better results, with follow-up periods as long as six years.16 -22 There seems little doubt that the Kendall program, and its subsequent versions, is efficacious and can be effective in reducing symptoms and the appearance of disorder.

In the second of the RCTs of prevention, Lowry-Webster et al15 used the Barrett et al18 program in a universal intervention with 594 children (aged 10–13 years) randomly allocated to CBT or to assessment only. The program was conducted by classroom teachers in normal class time. Post-treatment data showed a result similar to that in the targeted intervention study by Dadds et al;13 that is, the rates of remaining at risk in the intervention group were half those in the control group. Follow-up data are not yet available.

As with the interventions for preventing depressive disorders, the problem is not efficacy (CBT works), but effectiveness in routine practice. We are in the follow-up phase of an RCT of the Barrett et al18 program, now using school counsellors and teachers as therapists. We are already aware that, although school staff are prepared to conduct the prevention programs on one occasion, there is a reluctance to see this as part of their routine duties.

A recent meta-analysis of 10 well-designed studies of prevention programs in schools, aimed at reducing the prevalence and onset of drug use, found a positive effect.23 Botvin et al examined the effect at the six-year follow-up of a universal prevention program on tobacco, alcohol and marijuana use in 2800 early teens.24 For those who completed more than 60% of the program, significant reductions in prevalence were found across all three drug classes. In a sample of 12 000 students, Donaldson et al reduced the proportion of early teens using alcohol by 7%–10%25,26 (this is particularly important, as evidence suggests that the earlier adolescents begin using alcohol and other substances the more likely they are to misuse drugs in young adulthood27,28).

Both programs were designed to alter attitudes about the acceptability and prevalence of drug use and to teach skills to actively resist drug offers. Both included booster sessions in subsequent years, a factor that contributes to the general success of prevention interventions.23 However, universal prevention programs are more effective among non-users and experimenters than among frequent users or those who misuse.

Regular drug use in adolescents is not uncommon. In an Australian survey of 12–17-year-olds, 10% had used marijuana more than 40 times in the past year.29 For this subpopulation of regular users and misusers, a harm-reduction strategy is required. Such an approach seeks to reduce use and to minimise the risk of adverse consequences of use. Little formal evaluation of this approach has been conducted, but the few studies that exist show promise.30,31

One much-neglected solution is the use of large scale, simple, pragmatic, community randomised trials. These "effectiveness" trials can be seen as a step between the tightly controlled randomised "efficacy" trial (that is used to prove effect in a highly controlled environment) and the uncontrolled implementation of an intervention in a program setting.33 These trials aim to answer a simple question: What will the effect be when we apply this (proven) intervention into our community as a program?. The "community" in such trials is a geographically defined area such as a village, health district, or local council area, or a school, general practice, or healthcare service catchment area. Usually the intervention is applied at the level of the community — an entire town might receive a multifaceted intervention program against heart disease, or a school might receive a program designed to reduce depression or anxiety. The level of analysis is usually the level at which the intervention was applied, and so, although measures of effect are taken in individual people, the impact is aggregated at the community level and compared across communities to determine the effect. This makes intuitive sense, as what we are seeking to know is the effect at community level (we already know it works for individuals).

Community trials seek to have minimal influence on the community, other than through the intervention itself. The intent is to test the intervention in as "natural" a way as possible, and the research measurements themselves are designed to be as non-intrusive as possible (see Box).

Community randomised trials are not necessarily easy or cheap to do. Neither are they the only method available. Repeated time-series measures and before-and-after designs, with external controls, can also provide high quality evidence when designed and conducted with rigour. These are different to the more qualitative and process-type of evaluation that is typically done post-hoc. We would argue for the higher quality, pre-hoc, quantitative evaluation.

Few, if any, interventions that seek to promote mental health or prevent depression and anxiety have been subjected to large-scale community trials. They should be, and governments should have such trials done before spending millions of dollars on a national rollout of this type of intervention. Indeed, governments might consider supporting such trials as part of a national rollout. Rollout rarely occurs everywhere at once. How are initial sites chosen? Why not random allocation of communities? It is perhaps the fairest way. Communities could be randomised to the order in which they would receive the intervention, and those receiving the intervention later would act as controls for the early implementers. Fanciful? Not really: in Pakistan the government randomised health districts to the time when they would implement a new approach for tuberculosis control. This allowed researchers to set up a community randomised trial to test the effect of the new model in a highly rigorous way.35 The findings of the research raised several challenging issues that would otherwise have remained largely invisible. The result is, we hope, a better national program.

Australia has made a substantial contribution to the research evidence base regarding mental health prevention. Let us now be brave and smart enough to test some of these interventions in large-scale community trials, with funding support from the National Health and Medical Research Council (NHMRC) and the State and Commonwealth governments. Instead of insisting on, and funding post-hoc evaluations of, national programs, perhaps the funding could be provided up-front and used to support implementation through a community trial.

Such trials are not easy to do, but neither are they particularly hard. Done well, they need not be unaffordable, and obtaining high quality evidence for policy making is a sound investment.