Table 7 Two outcomes measures for comparison of cost-effectiveness: sources and quantification

Heroin abstinence

The mechanics of the transfer of the results from the effectiveness analysis over to the CEA for ‘heroin abstinence’ outcome followed three steps:

Step 1: The mixed effects regression analysis conducted on ‘heroin use’ produced individual-specific predicted values pertaining to the probability of ‘heroin use’ at each of the five time-points.

Step 2: The individual-specific predicted values of the probability of ‘heroin use’ at the final time-point was selected, and then manually reversed to become the probability of ‘heroin abstinence’. For example, if the probability of ‘heroin use’ of one participant was 0.78, the probability of ‘heroin abstinence’ of this participant would be 1–0.78 = 0.22.

Step 3: These individual-specific predicted values of the probability of ‘heroin abstinence’ were then used for estimating the ICER together with the cost data.

Drug-free days (over 36 months)

The mechanics of the transfer of the results from the effectiveness analysis over to the CEA for DFDs followed three steps:

Step 1: The mixed effects regression analysis conducted on DFDs (in the preceding 30 days)’ produced the individual-specific predicted values for each of the five time-points.

Step 2: Using these individual-specific predicted values of all five time-points to calculate the aggregated value over 36 months, using the formula described above.

Step 3: The individual-specific aggregated values for the DFDs (over 36 months)’ were then used for estimating the ICER together with the cost data.