Analytical approach:
A Markov decision model was constructed to estimate the long-term costs and outcomes, for a cohort of offenders, after two years of treatment. The time horizon was 25 years and the cycle length was one year. The authors stated that a service perspective was taken.
Effectiveness data:
The outcome measure was serious reoffending among life-sentence prisoners. A systematic review was conducted to identify studies reporting the rate of reconviction for serious offences, after specialist and mainstream incarceration. The annual rates of reconviction were between 2 and 5%. For the intervention, the annual reconviction rate was assumed to be 3%, based on offenders only being released if they were considered to have a low probability of reoffending. For the comparator, the annual rate was assumed to be 5%. All outcomes beyond one year were discounted at an annual rate of 3.5%.
Monetary benefit and utility valuations:
Reconvictions for serious offences were valued using estimates developed by the UK Home office. These included anticipated crime (defence expenditure, such as burglar alarms), the consequences of crime (physical and emotional impact on victims, victim services, and property), and responding to crime (criminal justice system and other services, such as treatment for victims of violent attacks). In the main analysis, it was assumed that reconvictions were for a mixture of offences, including serious wounding and sexual offences, and at least half of offences were homicides.
Measure of benefit:
The measures of benefit were the number of serious offences avoided, and the monetary value of these offences avoided.
Cost data:
The costs included those to the criminal justice system, and health and social services. The costs of one year of treatment were estimated, using the end-of-year financial statements from each DSPD programme site. Other costs were from various routine sources for unit cost data. All costs were reported in UK £, for the financial year 2005 to 2006. Costs beyond the first year were discounted at an annual rate of 3.5%.
Analysis of uncertainty:
Univariate sensitivity analyses were undertaken to assess the impact of plausible changes in key parameter values and the model structure on the results. The key parameters were the cost of the intervention, the duration of treatment, the probability of serious reoffending, and the discount rate. Probabilistic sensitivity analysis was conducted, using assigned distributions to assess the effect of combined parameter uncertainty on the results. For the net-benefit analysis, three scenario analyses were conducted, in which all reconvictions were serious wounding, all were serious sexual offences, or all were homicides. A cost-effectiveness acceptability curve was derived.