Hester van Eeren

| Chapter 2 2 | 24 Outcome measure: CAFY In health economic evaluations, cost-effectiveness is most commonly estimated in cost per quality-adjusted life year (QALY). However, as the predominant effect of behavioral interventions for criminal youths is the reduction of criminal activity (Erkenningscommissie Gedragsinterventies Jeugd, 2011) and thus is not directly or exclusively linked to physical health and absence of disease, the effect measure QALY seems inadequate to capture the full benefit of interventions in adolescent mental health (Sindelar, Jofre-Bonet, French, & McLellan, 2004). Therefore, a different outcome measure that sufficiently captures the goals of crime prevention and treatment was required. Considering the societal perspective of the policymaker, a broad outcome measure, directly linked to the goal of a reduction in criminal activity, was chosen. As a first step in this context, we chose the outcome measure of criminal activity free years (CAFYs), which can be used to determine the (incremental) costs per CAFY, i.e. the costs per criminal activity free year. Using CAFYs as the effect measure enables decisions based on a non-monetary value that is comparable between interventions and that properly reflects the goals of the Ministry of Justice while fitting into the health economic modelling approach. Existing examples of an effectiveness measure that resembles the use of the CAFY measure, is the use of days re-incarcerated (McCollister et al., 2004). As the model has two states defined as either being criminal or not being criminal, the transition from state A, criminal, to state B, not criminal, represents the rate of not being criminal after treatment. The transition of state B to state A on the other hand represents the rate becoming criminal after having been not criminal. It is assumed all youth enter the model as being criminal. The outcome of (incremental) costs per CAFY, was (as a first and rather simplified step) obtained by assigning different costs to individuals according to their current state, criminal state A or non criminal state B. Determining the net present value of the additional costs incurred in state A and state B over the full lifespan of subjects and dividing these by the amount of additional years the individual spends in the non criminal state B during his entire life (compared to TAU) yielded an estimate of incremental costs per CAFY. This process of calculating life-time costs and dividing these by life-time criminal-activity-free years was repeated 1000 times by means of simulation in order to reflect variability in input parameters. Model parameters: Transition probabilities Transition probabilities were dependent on the definition of the states reflecting the choice of outcome measure. In the current model, criminal behavior was chosen as most relevant outcome measure so that the states were defined as ‘criminal’ and ‘non criminal’ and transition probabilities between the states could be retrieved from literature. Several studies showed the effectiveness of FFT compared to TAU (French et al., 2008; Gordon, Graves, & Arbuthnot, 1995; Sexton & Turner, 2010; Sexton & Alexander, 2000, 2002). Yet, there is no consistent outcome regarding the effectiveness of FFT in comparison to TAU. The results based on adolescent recidivism derived from clinical trial findings reported by Sexton and Alexander (2000) were most applicable and comparable to the formulation of our model parameters and definition of the comparison group.