Hester van Eeren

Framework for modelling the cost-effectiveness of systemic interventions | 2 25 | So demonstrating the model, we used the effectiveness rates of that study (Sexton & Alexander, 2000). As the rate of recidivism based on the clinical trial reported in the study of Sexton and Alexander is 33 percent (Sexton & Alexander, 2000), we assumed this rate could be equal to the transition from state B to state A in the model and is therefore supposed to be equal to 33 percent. As the sum of all transition probabilities related to one state in the model sums up to 100 percent, the transition rate of state B to state B (individuals remaining in the non criminal state) is set at 67% (100% minus 33%). As for illustrative purposes we assumed here that the probability of individuals staying non-criminal (B to B) to be equal to the probability of becoming non-criminal (A to B),the transition from state A to state B, was fixed at 67 percent as well. Again subtracting this transition rate from 100% resulted in a probability of 33% for individuals remaining in the criminal state (A to A). Sexton and Alexander (2000) furthermore suggested that “FFT reduces recidivism and/or the onset of offending between 25 and 60 percent more effectively than other programs”. As TAU refers to a comparable treatment, we took the average of this range as a reasonable and illustrative estimate of the effectiveness of TAU. The model therefore was constructed under the illustrative assumption that FFT reduces criminal activity 42.5 percent more effectively than TAU. Transition probabilities were assumed to be fixed over the years, as no further long term effectiveness is known yet. Model parameters: Costs To fill in the cost parameters in the model the costs in the criminal state were retrieved from an ongoing trial of FFT (ZonMw, 2008). The volumes of costs in the non criminal state were derived from scaling volumes in the criminal state with a ratio of cost volumes of anti-social versus “normal” youths presented in a UK study on the financial costs of anti-social youths (Scott, Knapp, Henderson, & Maughan, 2001). Unit prices were taken from the Dutch manual for costing in economic evaluations (Hakkaart et al., 2010). In absence of Dutch unit costs, mean treatment costs of the interventions compared were derived from American costs presented in the study of Aos and colleagues (2004). These costs are not related to the states but depend on the intervention a youth received. Cost-effectiveness As the comparison of FFT with TAU in the current model is illustrative, the model results solely fulfil this objective. These illustrative cost-effectiveness results from the model point towards lower costs of FFT when compared to TAU. Taking the mean from the stochastic results, the number of CAFYs for FFT exceeds the number of CAFYs for TAU by 6.88 and the costs of FFT appear lower than TAU with incremental cost savings of 8,577EUR (Table 2), positioning the intervention in the South East quadrant of the cost-effectiveness plane (Figure 4). Incremental cost-effectiveness from the illustrative model data expressed in costs per CAFY amounts to cost savings of 1,246 EUR/CAFY. These exemplifying results suggest that FFT produces better effects at lower cost when compared to TAU.