Multidimensional Family Therapy in Cannabis Abusing Teens
INCANT was a phase III(b) randomised controlled effectiveness trial with an open-label, parallel group design, comparing MDFT with individual psychotherapy (IP) across sites in five Western European countries. Study sites started the 24-month recruitment phase between July 2006 and February 2007. Assessments were scheduled at baseline and at 3, 6, 9 and 12 months thereafter. INCANT was approved by an ethical board in each of the countries.
Eligible participants were teens between the ages of 13 and 18 years from both genders with a cannabis use disorder (dependence or abuse) established as having been present within the year prior to baseline. At least one parent had to take part in the treatment.
Adolescents were ineligible if they suffered from a current mental disorder or condition requiring inpatient treatment (e.g., psychosis, advanced eating disorder, suicidal ideation), or had a substance use disorder requiring maintenance treatment with methadone or buprenorphine. Cases were excluded if the adolescent and/or parent were unable to speak and read the local language.
We consecutively recruited 450 cases/families, with 60 to 120 cases/families per site depending on the local budget available.
The referral source for the adolescents varied. We distinguished two classes of referral. Self-Determined (SD) referral applied to adolescents who took the initiative to contact the treatment site on the advice (with no obligation) of relatives, friends, acquaintances, school, and occasionally on their own initiative. Externally Coerced (EC) referral applied to adolescents who felt they could not refuse treatment out of fear of sanctions, such as being kicked out of something (school, services, or programmes), being placed out of home, or being detained or otherwise being sanctioned by Justice authorities. This included referral by youth probation officers or appointed family guardians, public prosecutors, or court.
Sites were selected from addiction, youth and forensic care centres upon nomination by Government officials, after site visits by study staff, and on the basis of performance in a pilot study. The sites were (a) the outpatient cannabis clinic of the department of psychiatry of Brugmann University Hospital in Brussels, (b) Therapieladen in Berlin, (c) Centre Emergence in Paris with suburban CEDAT (Conseils Aide et Action contre le Toximanie) sub-sites in Mantes la Jolie and St Germain en Laye, (d) the twin sites of Parnassia Brijder (addiction care) and De Jutters (forensic care) in The Hague, and (e) Phénix in Geneva. All sites were youth oriented.
Concealed randomisation took place per site, using three stratification variables (gender, age and frequency of cannabis consumption), right after the case had been found eligible at the baseline assessment. For each stratum, the database computer generated 50 independent randomisations.
The database automatically assigned a code to each new case entered by a research assessor and informed her about the allocated treatment, independently from any trial staff.
Different therapists from the same site delivered either MDFT or IP. They were similar in age, gender, education and experience. Training procedures for MDFT therapists have been described before, and included a two weeks course in using the treatment manual, site visits, and active supervision during the study regarding session planning, case assessment and developing the treatment plan. Moreover, a sample of recordings of sessions were independently reviewed on measures of treatment adherence and competence. For IP, procedures for intervision and supervision were already in place, and where these were found wanting, they were upgraded to the level required for the study.
An average of two MDFT sessions per week was prescribed – in roughly equal proportion to be held with the adolescent, parent, and family (adolescent and parent together), respectively. Sessions could take place at the office of the therapist, the family's home, or any other location. IP was to last as long as MDFT (6 months), but with fewer sessions per week. Details on the actual treatment dose received have been reported before.
IP was individual counselling of the adolescent, which was treatment as usual across sites. IP was not standardised across sites; this would have required a revolution in deeply rooted treatment practices and would have aroused opposition from professional societies and other stakeholders – a challenge well beyond our capabilities. IP varied from full CBT in The Hague and Brussels to more elective approaches in the other countries. MDFT and IP were comparable in terms of motivational interviewing, intervision (peer consulting), administrative and referral procedures, and drug education.
Cannabis use was assessed with the Timeline Follow-Back method (TLFB), a well-validated self-report method to record frequency of cannabis use as adapted for adolescents. The TLFB registers daily cannabis use for the 90 days preceding the assessment, using a calendar and other memory prompts.
Also, urine samples were taken but these data were incomplete in Switzerland, where many clinicians regard urine testing as breaching the personal integrity of clients. This was already expected, and accepted as unchangeable, at the time the trial was designed. The laboratory results are not reported here.
Adolescents' internalising and externalising symptoms were recorded with the Youth Self Report (YSR). This instrument is reliable and valid across a variety of studies, populations, and languages (including Dutch, German and French). For the parents' view of the internalising and externalising symptoms of their children, the parent version of the YSR, called the CBCL (Child Behaviour Checklist), was used. The data used were from the parent most directly involved in raising the adolescent as established at baseline or from the parent volunteering to contribute to all follow-up assessments. The YSR and CBCL were scheduled at baseline and at six and twelve months.
We used the Anxiety/Depression, Withdrawn, and Somatic complaints sub-scales of the YSR and CBCL as components of the Internalising symptoms scale of the YSR and CBCL, which has been validated. A high score on this scale means that the adolescent (or in the CBCL, the parent) reported a high rate of internalising symptoms. For externalising symptoms, we used another validated scale, based on the sub-scales Delinquency and Aggressive behaviour. There is only one question on substance use in the YSR and CBCL among the 30 YSR externalising symptoms items, so overlap between our primary and secondary outcome measures was negligible.
Family conflict and cohesion were assessed with the respective sub-scales from the Family Environment Scale (FES), a widely used and well-validated self-report measure, which was completed by the teen. The FES was delivered at baseline and at six, nine and twelve months.
With a single exception, YSR, CBCL and FES analyses reported here are across-site. Information on outcomes per site has been published before.
We followed an intent-to-treat approach, with change in the YSR, CBCL, and FES outcomes (assessed at intake, 6, 9 [FES only], and 12 months) analysed using latent growth curve modelling (LGC;) conducted with the software Mplus (, version 6). Both intercept and slope were modelled separately for each outcome. We included site and referral source (self-directed or externally coerced referral) as covariates, based on INCANT's baseline findings, along with treatment condition. Site by treatment interactions were not statistically significant and therefore were omitted from the final models. Missing data were handled with full information maximum likelihood estimation, under the missing at random assumption.
A statistically significant (p < 0.05) slope parameter, as tested by the pseudo-z test, indicated the intervention was effective. For each slope intercept, we provide the 95% confidence interval (CI). In conditional models, MDFT was coded as 0 and IP as 1; thus, positive slope coefficients associated with treatment would indicate greater decreases in internalising and externalising problems for youth receiving MDFT. In addition, the magnitude of the coefficient indicated deviation from the mean slope associated with MDFT.
For comparing sites on continuous data we used analysis of variance (ANOVA).
Methods
Design
INCANT was a phase III(b) randomised controlled effectiveness trial with an open-label, parallel group design, comparing MDFT with individual psychotherapy (IP) across sites in five Western European countries. Study sites started the 24-month recruitment phase between July 2006 and February 2007. Assessments were scheduled at baseline and at 3, 6, 9 and 12 months thereafter. INCANT was approved by an ethical board in each of the countries.
Participants
Eligible participants were teens between the ages of 13 and 18 years from both genders with a cannabis use disorder (dependence or abuse) established as having been present within the year prior to baseline. At least one parent had to take part in the treatment.
Adolescents were ineligible if they suffered from a current mental disorder or condition requiring inpatient treatment (e.g., psychosis, advanced eating disorder, suicidal ideation), or had a substance use disorder requiring maintenance treatment with methadone or buprenorphine. Cases were excluded if the adolescent and/or parent were unable to speak and read the local language.
We consecutively recruited 450 cases/families, with 60 to 120 cases/families per site depending on the local budget available.
The referral source for the adolescents varied. We distinguished two classes of referral. Self-Determined (SD) referral applied to adolescents who took the initiative to contact the treatment site on the advice (with no obligation) of relatives, friends, acquaintances, school, and occasionally on their own initiative. Externally Coerced (EC) referral applied to adolescents who felt they could not refuse treatment out of fear of sanctions, such as being kicked out of something (school, services, or programmes), being placed out of home, or being detained or otherwise being sanctioned by Justice authorities. This included referral by youth probation officers or appointed family guardians, public prosecutors, or court.
Study Sites; Randomisation
Sites were selected from addiction, youth and forensic care centres upon nomination by Government officials, after site visits by study staff, and on the basis of performance in a pilot study. The sites were (a) the outpatient cannabis clinic of the department of psychiatry of Brugmann University Hospital in Brussels, (b) Therapieladen in Berlin, (c) Centre Emergence in Paris with suburban CEDAT (Conseils Aide et Action contre le Toximanie) sub-sites in Mantes la Jolie and St Germain en Laye, (d) the twin sites of Parnassia Brijder (addiction care) and De Jutters (forensic care) in The Hague, and (e) Phénix in Geneva. All sites were youth oriented.
Concealed randomisation took place per site, using three stratification variables (gender, age and frequency of cannabis consumption), right after the case had been found eligible at the baseline assessment. For each stratum, the database computer generated 50 independent randomisations.
The database automatically assigned a code to each new case entered by a research assessor and informed her about the allocated treatment, independently from any trial staff.
Therapists and Interventions
Different therapists from the same site delivered either MDFT or IP. They were similar in age, gender, education and experience. Training procedures for MDFT therapists have been described before, and included a two weeks course in using the treatment manual, site visits, and active supervision during the study regarding session planning, case assessment and developing the treatment plan. Moreover, a sample of recordings of sessions were independently reviewed on measures of treatment adherence and competence. For IP, procedures for intervision and supervision were already in place, and where these were found wanting, they were upgraded to the level required for the study.
An average of two MDFT sessions per week was prescribed – in roughly equal proportion to be held with the adolescent, parent, and family (adolescent and parent together), respectively. Sessions could take place at the office of the therapist, the family's home, or any other location. IP was to last as long as MDFT (6 months), but with fewer sessions per week. Details on the actual treatment dose received have been reported before.
IP was individual counselling of the adolescent, which was treatment as usual across sites. IP was not standardised across sites; this would have required a revolution in deeply rooted treatment practices and would have aroused opposition from professional societies and other stakeholders – a challenge well beyond our capabilities. IP varied from full CBT in The Hague and Brussels to more elective approaches in the other countries. MDFT and IP were comparable in terms of motivational interviewing, intervision (peer consulting), administrative and referral procedures, and drug education.
Outcome Measures
Cannabis use was assessed with the Timeline Follow-Back method (TLFB), a well-validated self-report method to record frequency of cannabis use as adapted for adolescents. The TLFB registers daily cannabis use for the 90 days preceding the assessment, using a calendar and other memory prompts.
Also, urine samples were taken but these data were incomplete in Switzerland, where many clinicians regard urine testing as breaching the personal integrity of clients. This was already expected, and accepted as unchangeable, at the time the trial was designed. The laboratory results are not reported here.
Adolescents' internalising and externalising symptoms were recorded with the Youth Self Report (YSR). This instrument is reliable and valid across a variety of studies, populations, and languages (including Dutch, German and French). For the parents' view of the internalising and externalising symptoms of their children, the parent version of the YSR, called the CBCL (Child Behaviour Checklist), was used. The data used were from the parent most directly involved in raising the adolescent as established at baseline or from the parent volunteering to contribute to all follow-up assessments. The YSR and CBCL were scheduled at baseline and at six and twelve months.
We used the Anxiety/Depression, Withdrawn, and Somatic complaints sub-scales of the YSR and CBCL as components of the Internalising symptoms scale of the YSR and CBCL, which has been validated. A high score on this scale means that the adolescent (or in the CBCL, the parent) reported a high rate of internalising symptoms. For externalising symptoms, we used another validated scale, based on the sub-scales Delinquency and Aggressive behaviour. There is only one question on substance use in the YSR and CBCL among the 30 YSR externalising symptoms items, so overlap between our primary and secondary outcome measures was negligible.
Family conflict and cohesion were assessed with the respective sub-scales from the Family Environment Scale (FES), a widely used and well-validated self-report measure, which was completed by the teen. The FES was delivered at baseline and at six, nine and twelve months.
With a single exception, YSR, CBCL and FES analyses reported here are across-site. Information on outcomes per site has been published before.
Analyses
We followed an intent-to-treat approach, with change in the YSR, CBCL, and FES outcomes (assessed at intake, 6, 9 [FES only], and 12 months) analysed using latent growth curve modelling (LGC;) conducted with the software Mplus (, version 6). Both intercept and slope were modelled separately for each outcome. We included site and referral source (self-directed or externally coerced referral) as covariates, based on INCANT's baseline findings, along with treatment condition. Site by treatment interactions were not statistically significant and therefore were omitted from the final models. Missing data were handled with full information maximum likelihood estimation, under the missing at random assumption.
A statistically significant (p < 0.05) slope parameter, as tested by the pseudo-z test, indicated the intervention was effective. For each slope intercept, we provide the 95% confidence interval (CI). In conditional models, MDFT was coded as 0 and IP as 1; thus, positive slope coefficients associated with treatment would indicate greater decreases in internalising and externalising problems for youth receiving MDFT. In addition, the magnitude of the coefficient indicated deviation from the mean slope associated with MDFT.
For comparing sites on continuous data we used analysis of variance (ANOVA).
SHARE
