Studying human development over time is tricky. Longitudinal studies are expensive, hard to plan, and hard to analyze. Study participants move on; sometimes researchers do, too. But in many cases a longitudinal study is the only way to find out what prevention researchers really want to know: will an intervention delivered to children still show benefits when those kids grow up?

A sterling example of the possibilities of a long-term study is the Montreal Longitudinal and Experimental Study (MLES). It started in 1984 with the goal of understanding and preventing delinquent behavior among boys. All too often, an evaluation of a behavior-improvement program like this one will assess effectiveness over a five-year period. The evaluators have to assume that intermediate outcomes – such as an assessment of behavior in elementary school – will be reflected in an end-point such as an adult criminal record.

However, such continuity is not certain. For evaluations with short follow-up periods, we simply don’t know if benefits shown in the short term persist. It leaves the question: what happens after?

The Montreal study provides rare insight into this question.

In 1984, 895 boys were recruited from kindergarten schools in deprived areas of Montreal, Canada. Baseline measures were taken of these boys’ social and disruptive behavior. The 250 who were found to be above the 70th percentile on the disruptiveness measures were defined as high risk of delinquency and school dropout and were randomized into intervention and control groups. The 645 lower-risk children were included as a “normative” comparison group.

The prevention program lasted two years. It gave training to parents (for example, teaching them to monitors their kids’ behavior, give positive reinforcement, and manage family crises), to students (in social skills), and to teachers.

Every spring, researchers asked the boys to answer a questionnaire about their behavior – for example, whether they stole, trespassed, took alcohol or drugs, vandalized things, or were in fights. These measures were taken each year until age 17, and then again at age 20. Parent and teacher reports were also used to rate the boys’ behavior from the start of the study to age 15. These intermediate outcomes were complemented by administrative data looking at criminal records and high school graduation.

Five to ten years into the program, the intermediate results, while promising, were not entirely consistent. For example, every year a higher proportion of boys participating in the intervention achieved “satisfactory school adjustment,” compared to high-risk boys not included in the intervention, but these results were only statistically significant for the years 1989 and 1990; in the other years, differences were small. There was no difference in boys’ self-reported arrests between the two groups from ages 11 to 13; from the age of 13 young people who received the intervention were arrested less then those who did not, but again, the difference was not significant.

Furthermore, it is likely these intermediate outcomes were also influenced by different drop-out rates between the intervention and control groups – which is another barrier to conducting good long term evaluations.

However, the picture painted of the impact of the Montreal program when using end-point outcomes was nothing short of startling. The two-year intervention, conducted with seven-to nine-year-old boys, resulted in 13% more of the intervention group graduating from high-school, significantly more then controls.

There was also a long-term effect on criminal records. Young adults who received the intervention were 11% less likely to have a criminal record by age 24, though this difference was just short of being statistically significant.

Perhaps more important, those who received the intervention caught up with their lower-risk peers on both graduation rate and the criminal record rate, suggesting the program was able to counterbalance the risk of early disruptive behavior.

While the long-term results are impressive, it is notable that all three groups’ outcomes were poor, because ‘the burden of other risk factors (i.e. low socio-economic status, inner-city residence) took its toll on the whole sample.”

Overall, the Montreal study shows the value of longitudinal studies: the only definitive way to evaluate the long-term impact of prevention programs.

References:
Boisjoli, R., Vitaro, F., Lacourse, E., Barker, E. & Tremblay, R. “Impact and clinical significance of a preventive intervention for disruptive boys: 15 year follow-up”, British Journal of Psychiatry, 191 (2007), 415-419.

Coalition for Evidence-Based Policy (2011) Montreal Prevention Program, The Rigorous Evidence Newsletter. Accessed from [http://coalition4evidence.org/wordpress/?page_id=923, 24 July 2011]

Tremblay, R. “The Montreal Longitudinal and Experimental Study: Tracing the Developmental Trajectories of Behavior Problems and Assessing their Prevention”, International Society of the Study of Behavioural Development 2010 Bulletin, 57(1) (2010), 21-24.