Hester van Eeren

| Chapter 2 2 | 18 al., 2002; French, McCollister, Sacks, et al., 2002). In addition, Aos and colleagues (2004) assessed costs and benefits from a taxpayer perspective. In health economic literature, cost-effectiveness analyses are preferably conducted from a societal perspective. Another difference between the two fields is, that in health economics sophisticated methodological guidelines for economic evaluations have been developed, while in the field of criminal justice such guidelines do not (yet) appear to exist. Furthermore, in health economic literature, cost-effectiveness or cost-utility analyses dominate (Soeteman & Busschbach, 2008). In the field of crime prevention and treatment, these analyses are limited. Nevertheless, McCollistar and colleagues (2003a; 2003b; 2004) and French and colleagues (2008) conducted various cost-effectiveness analyses related to substance abuse treatment, where the effectiveness is for example measured as days of re-incarceration (McCollister, French, Inciardi, et al., 2003; McCollister, French, Prendergast, et al., 2003; McCollister et al., 2004) or as a delinquency score (French et al., 2008). These studies show clearly the use of state of the art methods developed in the field of health care, applied in the field of crime prevention and treatment. On the other hand, these cost-effectiveness analyses were relatively conventional as parameter uncertainty was not captured in the model and long-term estimates were not taken into account. A common way to assess the cost-effectiveness in health care is the so- called decision analytic model (Briggs, Claxton, & Sculpher, 2006; Drummond, Sculpher, Torrance, O'Brien, & Stoddart, 2005). This approach provides a mathematical structure, synthesizing the evidence on costs and effects in a treated population under a variety of treatment options and makes the uncertainty around estimates visible. An additional advantage of this decision analytic modelling approach is that long-term effects can be modelled, even beyond the duration of the trial. Decision-analytic modelling and in particular inclusion of long-term effects may be especially relevant for interventions aiming to reduce criminal behavior. Several authors suggested that criminal behavior during adulthood tends to be preceded by behavioral disorders during childhood. Berger and Boendermaker (2003) stated that serious offenders often have a history of problematic behavior in their early years of life. Kim-Cohen and colleagues (2003) mentioned that most mental disorders in adults “...should be reframed as extensions of juvenile disorders”. This suggests that systemic interventions for juvenile disorders may reduce future criminal activity later on in life. Estimates of long-term effects are therefore essential to the analysis of these interventions. The current study aims to build a probabilistic decision analytic model like common models in health care for assessing interventions primarily aimed at crime prevention and treatment in youth care. In developing the model the following requirements had to be met: i. The model should be applicable to assess costs and effects of systemic interventions primarily aimed in reducing delinquent behavior; ii. The initial model should be fairly simple however easy to adjust to sophisticated details (i.e. severity of delinquency); iii. The model should be probabilistic, taking uncertainty into account; iv. The model should be suitable for long-term analysis;