Recidivism After Treatment in a Forensic Youth-Psychiatric Setting

Abstract

The aim of this study was to examine the effect of treatment characteristics on recidivism in a forensic youth-psychiatric outpatient clinic. The treatment offered comprised functional family therapy (FFT), individual cognitive behavioural therapy (CBT), or CBT in combination with parent training. Some of the youth additionally participated in aggression replacement training (ART). FFT and ART were implemented as a trial version, meaning that most therapists had not received formal training yet. Treatment characteristics related to recidivism were length of treatment, type of treatment, number of sessions, and the therapist. The longer the period of treatment and the greater the number of sessions, the higher the recidivism, even after controlling for risk of recidivism based on static risk factors. Juveniles who participated in ART reoffended more often than juveniles who had not participated in such training. Given the fact that FFT and ART were not well-implemented trial versions, it can be concluded that poorly implemented treatment leads to poor outcomes.

Keywords

effect of treatment characteristics recidivism FFT ART CBT

The main objective of the Dutch juvenile justice system is rehabilitation. Juvenile offenders are therefore often treated in outpatient settings instead of secure juvenile detention facilities. Outpatient settings in the Netherlands deliver various evidence-based behavioural interventions that target desistance from crime. These interventions include a number of individual interventions, such as cognitive behavioural therapy (CBT), and various types of skills training, as well as a number of system-oriented interventions, including parent training (PT), functional family therapy (FFT), and multisystemic therapy (MST), which are focussed on the juvenile as well as his or her environment. The aim of the present study was to evaluate factors that could have an effect on the outcomes of treatments in a Dutch forensic-psychiatric outpatient clinic. Treatments provided in this clinic are FFT and individual CBT, with or without PT. Some of the juveniles participated in aggression replacement training (ART) in addition to these interventions.

FFT has been developed in the United States and is aimed at juveniles between 12 and 18 years of age (and their family) who commit offenses on a regular basis and therefore come into contact with criminal law (Alexander & Parsons, 1982; Sexton & Alexander, 2004). FFT includes behavioural contracting, communication skills, specification of rules, and a token reinforcement system as techniques to improve communication. Research carried out in the United States into FFT (Alexander, Pugh, Parsons, & Sexton, 2000) has shown that it was effective in improving family functioning, family interactions, and communication in the family (Alexander, 1971; Alexander & Parsons, 1973; Friedman, 1989; Hansson, 1998; Klein, Alexander, & Parsons, 1977). However, a large-scale independent study into the effectiveness of FFT showed that juveniles treated with FFT did not reoffend less than juveniles from the control group (Barnoski, 2004). When a distinction was made between competent and less competent therapists, the recidivism rate was low among those treated by the competent therapists and high in the case of those treated by the less competent therapists. Therefore, recidivism is only reduced if FFT is given by competent therapists. If it is given by less competent therapists, recidivism increases rather than decreases. Another study of Sexton and Turner (2010) also showed that FFT can be effective in reducing recidivism, but only when therapists adhere to the treatment model.

Individual CBT is aimed at increasing positive behaviours and thoughts, decreasing negative behaviours and thoughts, and improving interpersonal skills (Lipsey, 2009). CBT is based on the fact that many young delinquents who repeatedly commit crimes see themselves as victims (Andrews et al., 1990; Lipsey & Wilson, 1998). CBT is a psychotherapeutic approach that addresses dysfunctional emotions, behaviours, and cognitions through a goal-oriented, systematic process (Lipsey & Landenberger, 2006). Various reviews have shown that CBT is effective in reducing recidivism (e.g., Andrews & Bonta, 2010; Lipsey, 1995; Lipsey & Wilson, 1998).

The aim of PT is to teach parents to help their child modify his or her behaviour. PT focuses on teaching parents a number of techniques based on social learning theories to help them change the problem behaviour, that is, how to reward desirable behaviour; how to give mild, appropriate punishments when misbehaviour occurs; and how to negotiate and make agreements with the child (Kazdin, 1987; Tarolla, Wagner, Rabinowitz, & Tubman, 2002). Research has shown that PT can lead to a decrease of recidivism (Behan & Carr, 2000; Brosnan & Carr, 2000; Kazdin, 1987). The effect of PT is greatest for the parents of juveniles in their early teenage years. The parents of older children proved less able to change their parenting style and were more likely to withdraw from training. Older children were more influenced than younger children by their peers and spent more time out of the home (Kazdin, 1997). PT is also less effective in families in which there are many risk factors associated with child dysfunction (Kazdin, 1987). In addition, it appears that PT is too difficult for some parents, for example, because they must be able to observe their child’s deviant behaviour and introduce procedures in the family (Kazdin, 1987).

Finally, ART is a group training in which three different skills are taught: anger management, social skills, and moral reasoning. Goldstein, Glick, and Gibbs (1998) developed ART for children and adolescents. Controlled studies have shown that it significantly improved the behaviour of aggressive or delinquent adolescents in various settings, provided the programme was implemented correctly (Barnoski, 2004). Barnoski (2004) found that the level of competence (proficiency) of the therapists was a very important success factor for the effect of ART. If ART is given by less competent therapists, recidivism increases rather than decreases.

The aforementioned studies have shown that the effectiveness of FFT, CBT, PT, and ART is dependent on certain circumstances, such as therapist competence and adherence to the treatment model. In this study, we examined the outcome of these interventions in a Dutch forensic-psychiatric outpatient setting. The aim of our study was to examine which (combinations of) treatment characteristics (type of intervention, total number of sessions, total duration of the intervention, number of sessions per month, therapist) lead to the highest reduction in recidivism. This knowledge can be used to improve treatment results.

Juveniles who received treatment for less than 2 months or had participated in less than six treatment sessions were defined as treatment dropouts. We first examined whether there were differences between these dropouts and the treatment group regarding static risk factors, such as gender, age, ethnicity, and number of previous offenses. The referral of juvenile delinquents to the different interventions was based mainly on the availability of therapists, and was thus quasi random. Therefore, we also examined differences in static risk factors between juveniles who were referred to the various forms of treatment because differences in static risk factors, such as gender, age, and ethnicity, may affect the outcomes.

We then examined differences in recidivism rates between the dropouts and the juveniles who were referred to the various forms of treatment. We also examined whether the level of recidivism for each type of treatment was related to the year in which the treatment was started because FFT and ART had just been introduced to the clinic. Because previous research has shown that the competence of the therapists (i.e., conducting the intervention precisely in accordance with the protocol) is crucial to the success of ART and FFT (Barnoski, 2004), we expect that recidivism is highest in the first years after introduction of ART and FFT, because in the first years, the therapists were less well trained and less experienced in conducting the intervention in accordance with the protocol.

Finally, we examined which treatment characteristics (total number of sessions, total duration of the therapy, number of sessions per month, therapist) had an impact on the level of recidivism, and which combinations of these treatment characteristics were associated with high or low recidivism. Because the associations between treatment characteristics and recidivism can depend on (static) risk factors for recidivism, rendering these associations spurious, we controlled for these risk factors. To summarize, in this study, we focussed on the following research questions:

Research Question 1: Are there differences in static risk factors between treatment group and dropouts and between the juveniles who participated in the various types of treatment?

Research Question 2: Are there differences in recidivism rates between treatment group and dropouts, and between the juveniles who participated in the various types of treatment?

Research Question 3: To what extent is the level of recidivism, for each type of treatment, related to the year in which the treatment was started?

Research Question 4: Which treatment characteristics have an impact on the level of recidivism, and which combinations of treatment characteristics are associated with high or low recidivism?

Method

Participants

The sample included 241 adolescents—207 boys (86%) and 34 girls (14%)—in the age range of 13 to 21 years (M_age = 16.7 years, SD = 1.84) who had been treated from 2002 to 2006 in the Bascule, a forensic-psychiatric outpatient clinic in Amsterdam. Juveniles who have been treated less than 2 months, or who have had less than six treatment sessions, were considered dropouts (n = 49). The dropouts were removed from the analysis of treatment characteristics (Research Questions 2, 4-6).

Treatment Setting

The Dutch forensic-psychiatric outpatient clinic was started in 2002. Juveniles who were referred to the clinic included juveniles who had been diagnosed with conduct disorder and who had come into contact with the justice system. These youth often have multiple psychiatric disorders. Besides conduct disorders, the following disorders occur: affective disorders, anxiety disorders, autism spectrum disorders, attention deficit hyperactivity disorder (ADHD), psychosis, alcohol and drugs abuse, and developing personality disorders.

During the first years, the organization was set up to implement evidence-based treatment in the clinic, and ART and FFT were implemented as trial versions, meaning that most therapists had not received formal training yet. The therapists used the treatment manuals, but they had not yet been properly trained to deliver the intervention as intended, and no treatment fidelity checks were used. The training of FFT started in September 2004 and the training of ART in 2006.

Measures

The demographic data and details of the treatment were obtained through file analysis. The following demographic variables were coded: gender, age at start of treatment, and ethnicity (Dutch or non-Dutch background). The following treatment characteristics were coded: type of treatment (FFT, CBT, or CBT with PT), additional participation in ART, total number of sessions, length of treatment (in months), average number of sessions per month, and therapist (there were 12 therapists who have treated the juveniles in the sample, and each therapist delivered several types of treatments).

Data regarding the criminal history were obtained from official records. The following variables regarding the criminal history were coded: age at first offense, total number of prior offenses, total number of prior misdemeanour offenses, total number of prior felony offenses, total number of prior violent offenses, and total number of prior property offenses. Both violent and property offenses can be felony offenses or misdemeanour offenses.

Outcome measure

Recidivism was obtained from official records and was defined as the occurrence of one or multiple new adjudications/convictions within 2 years after the start of the intervention. Therefore, the follow-up period was exactly 2 years for each youngster. All types of offenses were included, both felony and misdemeanour offenses. Recidivism was treated as a dichotomous variable (whether convicted for any new offense within a 2-year period).

Analyses

ANOVA was used to determine whether there were differences in the static risk factors between the treatment group (n = 192) and the dropouts (n = 49). The statistical power of an ANOVA test comparing two groups to detect a medium-sized effect with a sample size of n = 241 is 97% and should be considered adequate (Cohen, 1988). ANOVA was also used to determine differences in static risk factors between the different treatment groups (FFT, CBT, CBT with PT). The statistical power of an ANOVA test comparing three groups to detect a medium-sized effect with a sample size of n = 192 is 88% and should be considered adequate (Cohen, 1988).

Pearson correlations and chi-square tests were calculated to determine the strength of the relationship between the treatment characteristics and recidivism. A risk model was created to determine the risk of recidivism based on static risk factors. All background characteristics (gender, age at start of treatment, and ethnicity) and offense details (number and type of previous offenses, age at first offense) were included as independent variables in the model. Previous research has shown that these static risk factors are strong predictors of recidivism, and based on these variables, a good prediction of recidivism can be made (Van der Put, Deković, et al., 2011). Another risk assessment model was created to examine which combinations of treatment characteristics were associated with a high or low recidivism rate. This model was created for juveniles who had actually participated in treatment (n = 192), that is, excluding the dropouts (n = 49). All treatment characteristics (type of treatment, total number of sessions, total duration of the intervention, number of sessions per month, and therapist) were included as independent variables in this model.

Chi-square automatic interaction detector (CHAID) analyses were used to create the risk models. CHAID is highly appropriate for gaining insight into offender profiles with a high and low risk of reoffending, respectively, because it identifies groups of cases that share the same risk factors and also share the same risk of reoffending (Steadman et al., 2000; Thomas & Leese, 2003). In addition, CHAID’s advantages are that the results are visual and easy to interpret. The CHAID statistic is a classification tree method, which groups individuals into subsets with differing levels of risk on the basis of particular combinations of variables. This method focuses on interactions rather than on main effects in the data set being examined. The CHAID algorithm was used to assess the statistical significance of the bivariate association between each of the variables and recidivism until the most statistically significant value of χ² was reached. Once a variable was selected, the sample was divided (or partitioned) according to the values of that variable. This selection procedure was then repeated for each of the sample partitions, thus further partitioning the sample. The result of the partitioning process was to identify subgroups of cases sharing risk factor attributes that also exhibited high levels of homogeneity with regard to the level of risk of recidivism.

The predictive capacity of the models was assessed using the “area under the curve” (AUC) measure. The AUC indicates what percentage of correct classifications the instrument will yield overall (Hanley & McNeil, 1982). With a value of 0.50, the instrument does not predict better than chance. A value of 1.00 indicates a perfect positive prediction and a value of 0.00 indicates a perfect negative prediction. In general, an AUC value greater than 0.70 is considered acceptable for prediction and an AUC value greater than 0.75 is considered as high (Dolan & Doyle, 2000; Shapiro, 1999).

Results

Differences in Static Risk Factors Between Treatment Group and Dropouts

Table 1 shows the static risk factors (demographic and offense details) separately for the treatment group and the dropouts (juveniles who have been treated less than 2 months or who have had less than six treatment sessions).

Table 1.

Prevalence of Static Risk Factors Separately for the Treatment Group and the Dropouts.

	Treatment group (n = 192)	Dropouts (n = 49)	F
Male gender	85%	90%	0.77
Non-Dutch background (ethnic minority group)	48%	69%	7.38**
Age at the start of treatment (average)	17.0	16.6	1.72
Age at first offense (average)	14.4 years	14.2 years	1.19
Number of previous offenses (average)	3.9	4.8	2.73
Misdemeanour	1.0	1.2	1.11
Felony	2.9	3.6	2.24
Violent offenses	1.3	1.6	1.30
Property offenses	1.5	1.9	2.52

p < .01.

The groups differed significantly from each other with respect to the variable non-Dutch background. There were more juveniles with a non-Dutch background in the group of dropouts than in the treatment group.

Differences in Static Risk Factors Between the Different Treatment Groups

Table 2 shows the static risk factors separately for the different treatment groups. The groups differed significantly from each other with respect to the variables age at first offense and age at the start of treatment. Juveniles who were treated with CBT were older at the time of their first offense and at the start of the treatment than juveniles who were treated with FFT or CBT with PT.

Table 2.

Prevalence of Static Risk Factors Separately for the Different Treatment Groups.

	FFT (n = 55)	CBT (n = 87)	CBT + PT (n = 50)	F
Male gender	85%	85%	84%	0.02
Non-Dutch background (ethnic minority group)	47%	57%	48%	0.92
Age at the start of treatment (average)	16.3 years_a	17.0 years	16.1 years_a	5.20**
Age at first offense (average)	14.0 years_a	14.5 years	13.8 years_a	3.93*
Number of previous offenses (average)	4.1	3.8	3.9	0.15
Misdemeanour	1.2	1.0	0.8	0.95
Felony	3.0	2.9	3.1	0.15
Violent offense	1.2	1.4	1.3	0.21
Property offense	1.7	1.4	1.6	0.82

Note: FFT = functional family therapy; CBT = cognitive behavioural therapy; PT = parent training. Values sharing the same subscript do not differ significantly (p < .05).

p < .05. **p < .01.

Some of the youth participated in an Aggression replacement training (ART) in addition to their treatment with FFT, CBT, or CBT + PT. Table 3 shows the static risk factors separately for the juveniles who additionally participated in an ART training and those who did not additionally participate in an ART training. The groups differed significantly from each other on prior violent offenses. Juveniles who participated in an ART training in addition to their treatment with FFT, CBT, or CBT + PT had committed more violent offenses than juveniles who had not participated in an ART training in addition to their treatment with FFT, CBT, or CBT + PT.

Table 3.

Prevalence of Static Risk Factors Separately for Participants and Nonparticipants in ART.

	Additional ART (n = 27)	No additional ART (n = 165)	F
Male gender	78%	86%	1.24
Ethnic minority group	37%	55%	2.86
Age at the start of treatment (average)	16.3 years	16.6 years	0.69
Age at first offense (average)	14.0 years	14.2 years	0.27
Number of previous offenses (average)	4.5	3.9	0.99
Misdemeanour	1.2	1.0	0.93
Felony	3.2	2.9	0.43
Violent offense	2.3	1.1	11.68**
Property offense	1.1	1.6	2.35

Note: ART = aggression replacement training.

p < .01.

Differences in Recidivism Between the Dropouts and Treatment Group

Table 4 shows the recidivism rates (total and violent recidivism) within 2 years after the start of treatment, separately for the treatment group and the dropouts. There were no significant differences between the treatment group and the dropouts in the level of recidivism (total and violent recidivism).

Table 4.

Total and Violent Recidivism Separately for the Treatment Group and the Dropouts.

	Treatment group (n = 192)	Dropouts (n = 49)	F
Total recidivism	62.5%	55.1%	.89
Violent recidivism	32.3%	30.6%	.05

Because the only significant difference between the treatment group and the dropouts was that there were more juveniles with a non-Dutch background in the group of dropouts (see Table 1), we also examined whether the absence of a difference in recidivism continued to exist if we compared the two groups across ethnicity (see Table 5).

Table 5.

Total and Violent Recidivism for the Treatment Group and the Dropouts Separately for Juveniles With a Dutch and a Non-Dutch Background.

	Non-Dutch background (n = 126)			Dutch background (n = 115)
	Treatment group (n = 92)	Dropouts (n = 34)	F	Treatment group (n = 100)	Dropouts (n = 15)	F
Total recidivism	68.5%	50.0%	3.71*	57.0%	66.7%	0.50
Violent recidivism	38.0%	29.4%	0.80	27.0%	33.3%	0.26

p < .05.

In youth with a Dutch background, both total and violent recidivism were lower in the treatment group than in the dropouts group, but these differences were not significant. In youth with a non-Dutch background, however, both total and violent recidivism were higher in the treatment group than in the group of dropouts (only the difference in total recidivism was significant).

Differences in Recidivism Between the Different Treatment Groups

Table 6 shows the recidivism rates (total and violent recidivism) within 2 years after the start of treatment, separately for the different treatment groups. There were no significant differences between the treatment groups (FFT, CBT, CBT with PT) in the level of recidivism (total and violent recidivism).

Table 6.

Total Recidivism and Violent Recidivism for the Different Treatment Groups.

	Type of treatment				Additional ART
	FFT (n = 55)	CBT (n = 87)	CBT + PT (n = 50)	F	Yes (n = 27)	No (n = 165)	F
Total recidivism	68.3%	55.3%	67.3%	1.72	75.0%	59.9%	2.38
Violent recidivism	38.3%	28.7%	34.6%	0.80	53.6%	29.5%	6.68*

Note: ART = aggression replacement training; FFT = functional family therapy; CBT = cognitive behavioural therapy; PT = parent training.

p < .05.

Violent recidivism was significantly higher among juveniles who had participated in ART in addition to their treatment with FFT, CBT, or CBT + PT (54%) compared with juveniles who had not participated in ART in addition to their treatment with FFT, CBT, or CBT + PT (30%). Because these differences can partly be explained in terms of differences between the treatment groups (79% of the juveniles who additionally participated in ART had previously committed violent offenses, in comparison with 56% of the juveniles who did not additionally participate in ART next to their treatment), the same analysis was performed again, now only for the youth who had previously committed violent offenses. The results of this additional analysis showed that the violent recidivism rate among juveniles who had additionally participated in ART (54%) was still significantly higher (F = 4.4, p = .037) than the violent recidivism rate for juveniles who had not additionally taken part in ART (32%).

Recidivism Rate for Each Type of Treatment per Year

Table 7 shows the recidivism rates for each type of treatment by year of initiation of treatment (see Table 7). Recidivism was significantly lower in juveniles who started with FFT in 2004 than in juveniles who started with FFT before 2004.

Table 7.

Recidivism for Each Type of Treatment per Year Starting.

	2002		2003		2004		F
	n	Recidivism %	n	Recidivism %	n	Recidivism %	F
FFT	8	100_a	21	86_a	26	46	7.8**
CBT	35	49	28	54	24	71	1.5
CBT with PT	27	63	19	63	4	100	1.1
Additional ART	10	70	7	71	10	90	0.6
Total treatment group	70	60	68	66	54	61	0.3

Note: FFT = functional family therapy; CBT = cognitive behavioural therapy; PT = parent training; ART = aggression replacement training. Values sharing the same subscript do not differ significantly (p < .05).

p < .01.

Relationship Between Treatment Characteristics and Recidivism

Table 8 shows which treatment characteristics were related to recidivism. Pearson correlations were calculated for variables at interval level, and chi-square tests for the nominal variables. Treatment characteristics related with recidivism were total number of sessions (the greater the number of sessions, the higher the recidivism), length of treatment (the longer the period of treatment, the higher the recidivism), and the therapist (there were significant differences in recidivism rates between therapists; the rate varied between 45% and 83% per therapist).

Table 8.

Relationship Between Treatment Characteristics and Recidivism.

	χ²(df)	r
Total number of sessions		.18^*
Length of treatment		.18^*
Average number of sessions per month		−.08
Therapist	14.4(5)**

p < .05. **p < .01.

Combinations of Treatment Characteristics That Lead to High or Low Recidivism

To examine which combinations of treatment characteristics were associated with high or low recidivism, a CHAID analysis was performed in which all treatment characteristics (type of intervention, total number of sessions, total duration of intervention, number of sessions per month) were included as independent variables. Figure 1 shows the CHAID model, which had an AUC value of 0.67.

Figure 1.

Result of CHAID analysis based on treatment characteristics.

The most important predictor for recidivism was the treatment period. Based on this variable, the total group (100% with a risk of recidivism of 0.63) was split into two groups: a group having received treatment for 1 year or less (33% with a risk of recidivism of 0.47) and a group having received treatment for more than 1 year (67% with a risk of recidivism of 0.70). Then the group having received treatment for 1 year or less was divided again based on the number of sessions per month and the group having received treatment for more than 1 year was divided again based on type of treatment. The total group of youths was thus divided into four risk groups.

Juveniles who had received treatment for 1 year or less, attending more than two sessions per month, were least likely to reoffend (p = .34). If the average number of sessions per month was two or less, the risk of recidivism was considerably higher (p = .59). Juveniles who had been in treatment for longer than 1 year, and were treated with CBT, had a similar risk of recidivism (p = .59). The risk of recidivism was highest among juveniles who had been in treatment for longer than 1 year and received FFT or CBT with PT (p = .78).

Effect of Recidivism Risk on the Relationship Between Treatment Characteristics and Recidivism

The relationship between the treatment characteristics and recidivism might be spurious due to static risk factors. Therefore, we subsequently corrected for the risk of recidivism based on static risk factors. We first performed a CHAID analysis with all static risk factors (demographic and offense details) included as independent variables to calculate a risk score for each youth. Figure 2 shows the CHAID model, which has an AUC value of 0.71.

Figure 2.

Result of CHAID analysis based on static risk factors.

The most important predictor for recidivism was prior property offenses. Based on this particular variable, the total group (100% with a risk of recidivism of 0.61) was split into two groups: a group without prior property offenses (28% with a risk of recidivism of 0.40) and a group with prior property offenses (72% with a risk of recidivism of 0.69). Then the group without prior property offenses was divided again based on age at start of treatment, and the group with prior offenses was divided again based on age at start of treatment and gender. The total group of youths was thus divided into five risk groups. This model had an AUC of 0.72.

We then calculated partial correlations, corrected for the risk of recidivism based on static risk factors (see Table 9). The relationship between number of sessions and recidivism and the relationship between length of treatment and recidivism remained significant after controlling for recidivism risk.

Table 9.

Partial Correlations Between Treatment Characteristics and Recidivism, Controlling for Risk of Recidivism Based on Static Risk Factors.

	r
Total number of sessions	.16^*
Length of treatment	.17^*
Average number of sessions per month	−.07

p < .05.

We then examined whether the juveniles in the treatment segment with the highest recidivism rate (Figure 1) also had the highest risk of recidivism based on static risk factors, and whether the juveniles in the treatment segment with the lowest recidivism also had the lowest risk of recidivism based on static risk factors. Table 10 shows the recidivism risk based on static risk factors for each of the four treatment segments. The recidivism risk did not differ significantly between the treatment segments. Therefore, differences in recidivism between the treatment segments could not be explained by differences in static risk factors.

Table 10.

Risk of Recidivism Based on Static Risk Factors for Each Treatment Segment.

	Risk of recidivism
Longer than 12 months in treatment, family-oriented therapy	0.64
Longer than 12 months in treatment, individual cognitive behavioural therapy	0.59
Less than 12 months in treatment, up to two sessions per month	0.62
Less than 12 months in treatment, more than two sessions per month	0.57
F	2.7

Discussion

The aim of this study was to examine the effect of treatment characteristics on recidivism in a forensic youth-psychiatric outpatient clinic. The treatment comprised FFT, individual CBT, or CBT with PT. Some of the juveniles participated in ART in addition to these interventions.

The referral of juvenile delinquent to the interventions was based mainly on the availability of therapists, and may therefore be considered as quasi random. We therefore first examined whether there were differences in static risk factors between the juveniles who were referred to the different interventions. Juveniles who were referred to CBT were older at the time of their first offense and at the start of the treatment than juveniles who were referred to FFT or CBT with PT. Juveniles who were referred to an ART training had committed more violent offenses than juveniles who were not referred to an ART training. We also examined whether there were differences in static risk factors between the dropouts and the treatment group. The only significant difference between these groups was that there were more juveniles with a non-Dutch background in the group of dropouts than in the treatment group.

Second, we examined whether there were differences in 2-year recidivism rates between the different treatment groups, and between the treatment groups and the dropouts. There were no significant differences between the treatment groups (FFT, CBT, and CBT with PT) in the level of recidivism (total and violent recidivism). The violent recidivism rate was significantly higher among juveniles who had additionally taken part in aggression training compared with juveniles who had not taken part in such training. We examined whether this could be explained by differences in the number of previous violent offenses because juveniles receiving aggression training had committed more violent offenses than those not receiving aggression training. However, even if the figures were adjusted for the number of previous violent offenses, violent recidivism remained significantly higher among juveniles who had taken part in the aggression training. Another possible explanation is that the aggression training at the clinic was given to groups that entirely comprise deviant juveniles—Previous research has shown that group interventions in which the groups comprise only deviant juveniles can have a contrary effect on the behaviour of the participants (Dishion & Dodge, 2006; Dishion, McCord, & Poulin, 1999; Feldman, Caplinger, & Wodarski, 1983). The possible explanations for this include the presence of deviant role models, reward for and reinforcement of deviant behaviour, and negative labelling and stigmatisation. Another explanation can be found in the level of competence and proficiency of the therapists because the therapists in our study were not rigorously trained in ART. Barnoski (2004) found that recidivism is reduced only if aggression training is given by competent therapists. If it is given by less competent therapists, recidivism increases rather than decreases. The clinical explanation can be that noncompetent therapists do not counteract the “deviancy training.”

There also was no significant difference between the recidivism rate of the dropouts (55%) and the total treatment group (63%). Because the only significant difference between the dropouts and the treatment group was that there were more juveniles with a non-Dutch background in the group of dropouts, we separately examined differences between the dropouts and treatment group for youth with a Dutch background and for youth with a non-Dutch background. In youth with a Dutch background, both total and violent recidivism were lower in the treatment group than in the group of dropouts, but these differences were not significant. In youth with a non-Dutch background, however, both total and violent recidivism were higher in the treatment group than in the group of dropouts, where the difference was significant for total recidivism. It is difficult to explain why treatment could have an adverse effect in non-Dutch offenders (higher rates of drop out and recidivism). One reason could be that the treatment may not have been sufficiently culturally sensitive. It is possible that the family-oriented approach works less well with juveniles from ethnic minorities because a recent study showed that problems in the family domain were related to recidivism in ethnic majority youth, but not in ethnic minority youth (Van der Put, Stams, Dekovic, Hoeve, & Van der Laan, 2012). Therefore, it seems important to examine whether the treatment meets the specific requirements to effectively treat non-Dutch juvenile delinquents.

We then examined which treatment characteristics were related to recidivism. We found that the greater the number of sessions and the longer the period of treatment, the higher the recidivism rate was. The most obvious explanation for this is that the juveniles with the highest recidivism risk were treated the longest and received most sessions. The therapists decide how intensive and how long the treatment has to be, and it is therefore likely that therapists will continue the treatment if they believe that the risk of recidivism is high. However, even after controlling for risk of recidivism based on static risk factors, the relationship remained significant. Another possible explanation might be that the treatment period is extended when treatment does not have the desired effect. Previous intervention studies, however, have also shown that a longer treatment period and/or more sessions can lead to reduced effectiveness (Bakermans-Kranenburg, Van IJzendoorn, & Juffer, 2003; Dekovic et al., 2011; Lipsey & Wilson, 1998). Another explanation might be that the length of the treatment and the total number of sessions were not closely matched to the recidivism risk. Matching the intensity of an intervention to the recidivism risk, also referred to as the risk principle, is an important condition for the effectiveness of forensic interventions (Andrews & Dowden, 1999; Andrews et al., 1990; Lipsey, 1995). If the intensity of an intervention is not properly matched to the recidivism risk, its effect can even be counterproductive.

Another treatment characteristic related to recidivism was the therapist. There were striking differences between therapists in the level of recidivism. The level of recidivism varied between 45% and 83%. This means that it is important to monitor the effectiveness of each therapist and make clinical hypotheses about how these differences can be explained. It may be necessary to train and supervise the less competent therapists to make them more competent and adherent to evidence-based treatment models.

We then examined which combinations of treatment characteristics were associated with a high or low recidivism. Juvenile offenders least likely to reoffend were those who had received treatment 1 year or less, attending more than two sessions per month. The juveniles most likely to reoffend were those who had received treatment for more than 1 year, with a family-oriented therapy (FFT or CBT with PT). When the period of treatment exceeded 1 year, the recidivism after family-oriented therapy was significantly higher than the recidivism after individual CBT. This can be partly attributed to differences in the static risk factors between the participant groups. Juveniles who were treated with CBT were older at the time of their first offense and at the start of the treatment than juveniles who were treated with FFT or CBT with PT.

It is likely that the high recidivism rates among juveniles receiving FFT may be attributable to implementation problems, as FFT had just been introduced when this study started. The recidivism was significantly lower for juveniles who started FFT 2 years after the introduction, when the therapists had been properly trained, than for those who started FFT just after it was introduced, when therapists had not yet been properly trained. These results suggest that first experience had to be gained with “adherence” (i.e., conducting the intervention precisely in accordance with the protocol), which is apparently crucial to the success of family therapy (Alexander et al., 2000; Sexton & Turner, 2010). FFT was developed as a short-term intervention compromising an average of 20 sessions, no more than 30 sessions, and lasting 3 to 6 months (Alexander & Parson, 1982). Nevertheless, during the pilot years of the clinic, FFT had an average length of 20 months, and the average number of sessions was 30. The number of sessions was, with an average of 1.5 sessions per month, considerably less than the intended 4 sessions per month. The intensity of FFT was thus much lower than it was intended, which may have influenced the effectiveness of FFT.

Another possible explanation for the relatively high recidivism rate among juveniles who were treated with a family-oriented therapy is that juveniles who are on average age 16 might be already “too old” for this form of therapy. Recent research has shown that the impact of the family on recidivism decreases sharply as juveniles get older (Van der Put et al., 2010; Van der Put, Dekovic, et al., 2011). Problems in the family domain are still the main predictors for recidivism in 12-year-olds, but the importance decreases sharply after that age. For juveniles of 14 years of age and above, risk factors in the domains of school, relationships, and attitude have a substantially higher impact than risk factors in the family domain. Research into the effect of PT also shows that its effects are greatest for the parents of juveniles in their early teenage years (Kazdin, 1997). Given that juveniles older than 14 years accounted for 90% of the research group, it is possible that focusing on the family is not a fruitful approach.

This study has a number of limitations. First, the juveniles were not randomly assigned to the various types of treatment. Although it was not the purpose of this study to demonstrate treatment effects, random assignment of juveniles to the various treatments is also preferable from the perspective of gaining insight into moderators. We found no significant differences between the various treatment groups in most of the static risk factors, but there could be differences between the groups on variables not captured by the data collection that would nonetheless affect the outcomes of the study. A second limitation is that a number of important independent variables (e.g., psychopathology and problems in the family situation) could not be taken into account because there was insufficient information on them in the records. This means that apart from demographic and offense details, no dynamic risk and protective factors could be examined. A third limitation was the impossibility to filter out last minute cancelled appointments from the system. Therefore, we could not distinguish between an appointment that was cancelled last minute and a nonmissed appointment. As a result, some missed appointments were analyzed as treatment sessions. A fourth limitation of this study is that it did not look specifically at how soon the juveniles reoffended and how serious the reoffenses were. Finally, it should be noted that our results are based on a single study, conducted in a single treatment setting. The recommendation is, therefore, to replicate this study in other treatment facilities to obtain greater insight into moderators of treatment effectiveness within populations of juvenile offenders.

One of the strong points of this study is its ecological validity, as it was carried out in a forensic-psychiatry practice setting (“real-world” setting; Barkham & Mellor-Clark, 2000; Cape & Barkham, 2002). The findings of this study can therefore be applied directly in practice to improve the effectiveness of treatment. The most important findings of this study were that treatment had a negative effect on juveniles (a) when treatment was provided by less well-trained therapists, (b) when the intensity of the treatment was not well matched to the recidivism risk of the juveniles, (c) when the therapists did not adhere well to the treatment model (treatment period was too long and not sufficiently intensive), and (d) when the referral of youth to the various forms of treatment was based on the availability of therapists instead of criminogenic needs of the juveniles. Effectiveness can be enhanced by improving the aforementioned issues, for example, by matching the length of treatment and the number of sessions more closely to the recidivism risk. Current research shows that a longer treatment period (longer than 12 months) is linked to increased recidivism, which cannot be explained by a high recidivism risk. This indicates that treatment effectiveness can be improved by limiting the treatment period (12 months or less) and increasing the frequency of sessions (more than two sessions per month).

Effectiveness of the treatment offered can possibly also be increased by properly referring juveniles to the appropriate type of treatment. This requires a reliable and valid assessment of the criminogenic needs. Structured assessment instruments are necessary for this purpose. At the time of the present study, such instruments were not available in the Dutch juvenile justice system. Meanwhile, a new set of instruments (Landelijk Instrumentarium Jeugdstrafrechtketen [LIJ]; Van der Put, Stams, et al., 2011) has been developed geared to the various phases of the juvenile justice system. The LIJ aims to improve the assessment of psychosocial and recidivism risks and referral to appropriate evidence-based interventions.

In this study, FFT and ART were still implemented poorly. In FFT and ART, the therapists were not properly trained and the therapists did not adhere to the principles of the intervention. For example, the treatment period was too long in FFT, and the training in ART was given to groups entirely comprising deviant juveniles. Lipsey (2009) also found in a meta-analytic study that quality of implementation was one of the three major correlates of programme effectiveness. In conclusion, implementing treatments as trial versions, in which therapists are poorly trained, can lead to an increase rather than a decrease of recidivism. Therefore, an important lesson to be learned from this study is that poorly implemented treatment leads to poor outcomes.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interests with respect to the authorship and/or publication of this article.

Funding

The author(s) received no financial support for the research and/or authorship of this article.

References

Alexander

J. F.

(1971). Evaluation summary: Family groups treatment program. Report to Juvenile Court, District 1, State of Utah, Salt Lake City.

Alexander

J. F.

Parsons

B. V.

(1973). Short term behavioral intervention with delinquent families: Impact on family process and recidivism. Journal of Abnormal Psychology, 81, 219-225.

Alexander

J. F.

Parsons

B. V.

(1982). Functional family therapy: Principles and procedures. Carmel, CA: Brooks/Cole.

Alexander

J. F.

Pugh

Parsons

Sexton

(2000). Book three: Functional family therapy. In Elliot

D. S.

(Ed.), Blueprints for violence prevention (pp. 117-140). Golden, CO: Venture Publishing.

Andrews

D. A.

Bonta

(2010). Rehabilitating criminal justice policy and practice. Psychology, Public Policy, and Law, 16, 39-55.

Andrews

D. A.

Dowden

(1999). A meta-analytic investigation into effective correctional intervention for female offenders. Forum on Corrections Research, 11, 18-21.

Andrews

D. A.

Zinger

Hoge

R. D.

Bonta

Gendreau

Cullen

F. T.

(1990). Does correctional treatment work? A clinically relevant and psychologically informed meta-analysis. Criminology, 28, 369-404.

Bakermans-Kranenburg

M. J.

Van IJzendoorn

M. H.

Juffer

(2003). Less is more: Meta-analyses of sensitivity and attachment interventions in early childhood. Psychological Bulletin, 129, 195-215.

Barkham

Mellor-Clark

(2000). Rigour and relevance: Practice based evidence in the psychological therapies. In Rowland

Goss

(Eds.), Evidence based counselling and psychological therapies: Research and applications (pp. 127-144). London, England: Routledge.

10.

Barnoski

(2004). Outcome evaluation of Washington state’s research-based programs for juvenile offenders. Olympia: Washington State Institute for Public Policy.

11.

Behan

Carr

(2000). Oppositional defiant disorder. In Carr

(Ed.), What works with children and adolescents? A critical review of psychological interventions with children, adolescents and their families (pp. 102-130). London, England: Routledge.

12.

Brosnan

Carr

(2000). Adolescent conduct problems. In Carr

(Ed.), What works with children and adolescents? A critical review of psychological interventions with children, adolescents and their families (pp. 131-154). London, England: Routledge.

13.

Cape

Barkham

(2002). Practice improvement methods: Conceptual base, evidence based research and practice based recommendations. British Journal of Clinical Psychology, 41, 285-307.

14.

Cohen

(1988). Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum.

15.

Dekovic

Slagt

M. I.

Asscher

J. J.

Boendermaker

Eichelsheim

V. I.

Prinzie

(2011). Effects of early prevention programs on adult criminal offending: A meta-analysis. Clinical Psychology Review, 31, 532-544.

16.

Dishion

T. J.

Dodge

K. A.

(2006). Deviant peer contagion in interventions and programs. An ecological framework for understanding influence mechanisms. In Dodge

K. A.

Dishion

T. J.

Lansford

J. E.

(Eds.), Deviant peer influences in programs for youth (pp. 14-43). New York, NY: Guilford.

17.

Dishion

T. J.

McCord

Poulin

(1999). When interventions harm: Peer groups and problem behaviour. American Psychologist, 54, 755-764.

18.

Dolan

Doyle

(2000). Violence risk prediction: Clinical and actuarial measures and the role of Psychopathy Checklist. British Journal of Psychiatry, 177, 303-311.

19.

Feldman

R. A.

Caplinger

T. E.

Wodarski

J. S.

(1983). The St. Louis conundrum: The effective treatment of antisocial youth. Englewood Cliffs, NJ: Prentice Hall.

20.

Friedman

A. S.

(1989). Family therapy vs. parent groups: Effects on adolescent drug abusers. American Journal of Family Therapy, 17, 335-347.

21.

Goldstein

A. P.

Glick

Gibbs

J. C.

(1998). Aggression replacement training: A comprehensive intervention for aggressive youth. Champaign, IL: Research Press.

22.

Hanley

J. A.

McNeil

B. J.

(1982). The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology, 143, 29-36.

23.

Hansson

(1998, February). Functional family therapy replication in Sweden: A cost-benefit analysis of in-home family therapy vs. probation for juvenile delinquents. Paper presented at the annual Banff Conference on Behavioral Sciences, Banff, Alberta, Canada.

24.

Kazdin

A. E.

(1987). Treatment of antisocial behaviour in children: Current status and future directions. Psychological Bulletin, 102, 187-189.

25.

Kazdin

A. E.

(1997). Practitioner review: Psychosocial treatments for conduct disorder in children. Journal of Child Psychology and Psychiatry, 62, 161-182.

26.

Klein

N. C.

Alexander

J. F.

Parsons

B. V.

(1977). Impact of family systems intervention on recidivism and sibling delinquency: A model of primary prevention and program evaluation. Journal of Consulting and Clinical Psychology, 45, 469-474.

27.

Lipsey

M. W.

(1995). What do we learn from 400 research studies on the effectiveness of treatment with juvenile delinquents? In McGuire

(Ed.), What works: Reducing reoffending. Guidelines from research and practice (pp. 63-78). Chichester, UK: Wiley.

28.

Lipsey

M. W.

(2009). The primary factors that characterize effective interventions with juvenile offenders: A meta-analytic overview. Victims and Offenders, 4, 124-147.

29.

Lipsey

M. W.

Landenberger

N. A.

(2006). Cognitive-behavioral interventions. In Welsh

B. C.

Farrington

D. P.

(Eds.), Preventing crime. What works for children, offenders, victims and places (pp. 83-127). New York, NY: Springer.

30.

Lipsey

M. W.

Wilson

D. B.

(1998). Effective intervention for serious juvenile offenders. A synthesis of research. In Loeber

Farrington

D. P.

(Eds.), Serious and violent juvenile offenders. Risk factors and successful interventions (pp. 313-338). London, England: Sage.

31.

Lösel

(1995). The efficacy of correctional treatment. A review and synthesis of meta-evaluations. In McGuire

(Ed.), What works: Reducing reoffending, guidelines from research and practice (pp. 79-111). Chichester, UK: Wiley.

32.

Schwalbe

C. S.

(2007). Risk assessment for juvenile justice: A meta-analysis. Law and Human Behavior, 31, 449-462.

33.

Sexton

Alexander

J. F.

(2004). Functional family therapy: Clinical training manual. Seattle, WA: Annie E. Casey Foundation.

34.

Sexton

Turner

C. W.

(2010). The effectiveness of functional family therapy for youth with behavioral problems in a community practice setting. Journal of Family Psychology, 24, 339-348.

35.

Shapiro

D. E.

(1999). The interpretation of diagnostics tests. Statistical Methods in Medical Research, 8, 113-134.

36.

Steadman

H. J.

Silver

Monahan

Appelbaum

P. S.

Robbins

P. C.

Mulvey

E. P.

. . .Banks

(2000). A classification tree approach to the development of actuarial violence risk assessment tools. Law and Human Behavior, 24, 83-100.

37.

Tarolla

S. M.

Wagner

E. F.

Rabinowitz

Tubman

J. G.

(2002). Understanding and treating juvenile offenders: A review of current knowledge and future directions. Aggression and Violent Behavior, 7, 125-143.

38.

Thomas

S. D.

Leese

(2003). Invited editorial. A green-fingered approach can improve the clinical utility of violence risk assessment tools. Criminal Behaviour and Mental Health, 13, 153-158.

39.

Van der Laan

P. H.

Slotboom

A. M.

Stams

G. J. J. M.

(2010). Wat werkt? In van Koppen

P. J.

Merkelbach

Jelicic

de Keijser

J. W.

(Eds.), Reizen met mijn rechter: Psychologie van het recht (pp. 987-1002). Deventer, Netherlands: Kluwer.

40.

Van der Put

C. E.

Dekovic

Stams

G. J. J. M.

Van der Laan

P. H.

Hoeve

Van Amelsfort

(2011). Changes in risk factors during adolescence: Implications for risk assessment. Criminal Justice and Behavior, 38, 248-262.

41.

Van der Put

C. E.

Stams

G. J. J. M.

Dekovic

Hoeve

Van der Laan

P. H.

(2012). Ethnic differences in the prevalence and impact of risk factors for recidivism. Manuscript submitted for publication.

42.

Van der Put

C. E.

Stams

G. J. J. M.

Dekovic

Hoeve

Van der Laan

P. H.

Spanjaard

Barnoski

(2010). Criminogenic needs during adolescence: Changes in the relative importance of dynamic risk factors for recidivism during adolescence. International Journal of Offender Therapy and Comparative Criminology. Advance online publication. doi:10.1177/0306624X11398462

43.

Van der Put

C. E.

Stams

G. J. J. M.

Spanjaard

H. J. M.

Van Domburgh

Doreleijers

T. A. H.

Lodewijks

H. P. B.

. . . Bolt

R. B.

(2011). Risk and needs assessment in the Netherlands: The development of a new assessment device for juvenile delinquents in The Netherlands. Manuscript submitted for publication.

44.

Wartna

B. S. J.

Tollenaar

Blom

(2005). Recidive 1997. Een cijfermatig overzicht van de strafrechtelijke recidive van volwassen en jeugdige daders. Den Haag, Netherlands: WODC.

45.

Weisz

J. R.

Jensen Doss

Hawley

K. M.

(2005). Youth psychotherapy outcome research: A review of critique of the evidence base. Annual Reviews of Psychology, 56, 337-363.