MDFT (Multidimensional Family Therapy) is a family based outpatient treatment programme for adolescent problem behaviour. MDFT has been found effective in the USA in adolescent samples differing in severity and treatment delivery settings. On request of five governments (Belgium, France, Germany, the Netherlands, and Switzerland), MDFT has now been tested in the joint INCANT trial (International Cannabis Need of Treatment) for applicability in Western Europe. In each of the five countries, study participants were recruited from the local population of youth seeking or guided to treatment for, among other things, cannabis use disorder. There is little information in the literature if these populations are comparable between sites/countries or not. Therefore, we examined if the study samples enrolled in the five countries differed in baseline characteristics regarding demographics, clinical profile, and treatment delivery setting.

INCANT was a multicentre phase III(b) randomized controlled trial with an open-label, parallel group design. It compared MDFT with treatment as usual (TAU) at and across sites in Berlin, Brussels, Geneva, The Hague and Paris.

Participants of INCANT were adolescents of either sex, from 13 through 18 years of age, with a cannabis use disorder (dependence or abuse), and at least one parent willing to take part in the treatment. In total, 450 cases/families were randomized (concealed) into INCANT.

We collected data about adolescent and family demographics (age, gender, family composition, school, work, friends, and leisure time). In addition, we gathered data about problem behaviour (substance use, alcohol and cannabis use disorders, delinquency, psychiatric co-morbidity).

There were no major differences on any of these measures between the treatment conditions (MDFT and TAU) for any of the sites. However, there were cross-site differences on many variables. Most of these could be explained by variations in treatment culture, as reflected by referral policy, i.e., participants' referral source. We distinguished 'self-determined' referral (common in Brussels and Paris) and referral with some authority-related 'external' coercion (common in Geneva and The Hague). The two referral types were more equally divided in Berlin. Many cross-site baseline differences disappeared when we took referral source into account, but not all.

A multisite trial has the advantage of being efficient, but it also carries risks, the most important one being lack of equivalence between local study populations. Our site populations differed in many respects. This is not a problem for analyses and interpretations if the differences somehow can be accounted for. To a major extent, this appeared possible in INCANT. The most important factor underlying the cross-site variations in baseline characteristics was referral source. Correcting for referral source made most differences disappear. Therefore, we will use referral source as a covariate accounting for site differences in future INCANT outcome analyses.