Hester van Eeren

| Chapter 3 3 | 44 Discussion While cost-effectiveness analyses are increasingly being used in the field of crime prevention, the value of further research has not yet been estimated for comparison between interventions aimed at reducing juvenile delinquency. An earlier developed cost-effectiveness model was used to estimate this value of further research. This study demonstrated that it was feasible to estimate the value of conducting further research in this context, using a value of information framework common in health economic evaluations. The results can be interpreted as similar to cost/QALY (Quality-Adjusted Life Year) studies in health care evaluation. In this value of information analysis, the results indicated the parameters for which further research was valuable. Our findings show particular uncertainty in three groups of parameters: the transition probabilities of the Course House and of FFT, and to a lesser extent, the intervention costs of the Course House and the direct non health-care costs in both model states. Performing additional research in the suggested fields can reduce parameter uncertainty, and hence, can reduce decision uncertainty. Therefore, the results of a value of information analysis can prioritize further research to optimize the final reimbursement decision, thereby increasing the probability that adolescents will be assigned to the intervention that is cost-effective, compared with the alternative. Given this information, future interventions could be reimbursed (or not), and they could also be approved ‘only in research’ (OIR) (i.e. further research is required before the intervention can be approved) or ‘approved with research’ (AWR) (i.e. research can be conducted while the intervention is approved) (Claxton et al., 2012; McKenna & Claxton, 2011). For example, from this study we can conclude that given a WTP of €40,000 per CAFY, the Course House could be ‘approved with research’. The Course House would then be reimbursed while further research would be required, for example on the effectiveness of the Course House. Current practice in adolescent care in the Netherlands illustrates this approval condition: the Netherlands Youth Institute identifies effective youth interventions, while still conducting research on the effectiveness of some of these interventions (Netherlands Youth Institute, 2014). However, approval might lead to irrecoverable costs when the approval is revised due to subsequent research revealing that the Course House was not as effective as expected. Then, approval ‘only in research’ might be preferred, because commitment to future costs is avoided until the results of further research are known. Approval might even be dependent on any change in the effective price of an intervention (Claxton et al., 2012; Walker, Sculpher, Claxton, & Palmer, 2012). This study was a first attempt to apply a value of information framework to the field of crime prevention and treatment of juvenile delinquents. Therefore, some considerations should be kept in mind. The value of information analysis estimates the monetary value of eliminating all or part of the parameter uncertainty of the presented model. However, two other sources of uncertainty can influence the results: structural and methodological uncertainty (Bojke, Claxton, Sculpher, & Palmer, 2009; Briggs, 2000). Structural uncertainty relates to structural aspects of the model (Bilcke, Beutels, Brisson, & Jit, 2011; Bojke et al., 2009; Haij Ali Afzali & Karnon, 2015), such as the conceptual framework or the transitions between the model states (Haij Ali Afzali &