Figure 2.

Power curve of the required sample-size per treatment arm. The curve is represented as a function of willingness-to-pay (WTP) for a range of increases in costs (delta Costs) when a treatment strategy is not only more effective but also more costly. The presented +sample-size is unadjusted for intra-cluster correlation. A minimally important change in patient utility (EQ-5D) has been defined as 0.074 point. With 150 patients per treatment strategy we are able to detect at least a change of 0.06 points by net health benefit analysis between the arms with a SD = 0.175 EQ-5D points (baseline data SMASHING-project: SD = 0.17) and a SD of €1000 for costs (SD = €816, usual care strategy) and an increase in costs of €250 (delta Costs) when a treatment strategy is not only more effective but also more costly, for a willingness-to-pay (WTP) of €30K (alpha = 0.05, one-sided, beta = 0.20, one-sided, rho costs-effects = 0). With 40 clusters (general practices) per arm and assuming an intra-cluster correlation of 0.01, 0.07 and 0.11 the number of patients per cluster is 4, 5, and 6, and the total number of patients is 480, 600 and 720, respectively.