Hester van Eeren

Framework for modelling the cost-effectiveness of systemic interventions | 2 27 | Discussion and Conclusions This study created a framework for the evaluation of interventions aimed at reducing criminal activity in delinquent youth. A probabilistic Markov model approach was constructed allowing the assessment of the incremental cost-effectiveness of two systemic interventions. For illustrative purposes, the interventions considered were FFT and TAU. As the comparison of FFT with TAU in the current model is solely an example to demonstrate model functioning, the model results are illustrative in absence of empirical data. As a first step to come to suitable outcome measures in this field, we introduced the outcome measure of Criminal Activity Free Years (CAFY) in a probabilistic decision analytic model. The presented methodology may provide a basis for further development of the model and outcome measures and, ultimately, decision-making by both Ministries of Justice and, in particular, Health. Policymakers may compare cost and effects between different types of interventions aiming to reduce delinquency among youth. An advantage of using decision analytic models is that this approach enables calculation of hypothetical scenarios. Hence, questions of policymakers, for example on differences in cost-effectiveness within subgroups of youth or on the optimal age for intervention may be answered. Moreover, the decision uncertainty is represented in the model results by taking into account the uncertainty surrounding the input parameters of the model. The current study showed that it was feasible to apply health economic methodology to assess interventions aimed at reducing delinquency rates. The approach was developed to be consistent with health economic guidelines. To our knowledge, this was the first economic evaluation using decision-analytic modelling in the evaluation of systemic interventions for crime prevention and treatment. However, a number of important questions remain. First of all, the outcome measure presented here is clearly sector-specific. While this enables choosing between interventions with similar aims, it does not directly allow comparisons with other interventions. This problem is not unique for this context. For instance, interventions in elderly care or social care may not be primarily aimed at producing health as well. Outcome measures such as the OPUS and ICECAP have been proposed as better capturing the benefits of such care (Coast et al., 2008; Ryan, Netten, Skatun, & Smith, 2006). This does raise the question, however, of how to trade-off between interventions when their aim is not similar and when different outcome measures were used to assess cost-effectiveness. This seems to be an important area for future research. Secondly, we proposed the measure of CAFY as a first step to demonstrate how interventions aimed to reduce delinquency could be evaluated within a probabilistic decision model. If such interventions were to be evaluated more systematically using methodology like the one presented here, clearly, the outcome measure deserves more attention. The outcome measure of the CAFY is a very simple and crude one. One could compare it to ‘natural units’ used in cost-effectiveness analysis like gained life years and event free life years. An important problem with these measures and the CAFY is that they do not reflect the seriousness of the events (e.g. living in a poor or good health state or, in this case, engaging in many and severe criminal activities or a few minor felonies). However, the definition of criminal activity free could be based on different