Jeroen van de Pol

93 4 Cost-utility and cost-effectiveness analysis of a clinical medication review Valuation Healthcare utilization was valued according to guidelines for economic evaluation in healthcare in the Netherlands [29]. Informal care was valued according to iMTA (Medical Technology Assessment) at €14 per hour (2014 prices) and was indexed to 2017 prices. The amount of time for informal care was maximized at 16 hours per day. Drug costs were presented in 2017 euros. Prices from previous years were updated according to the Dutch consumer price index [30]. The costs of the intervention were calculated by multiplying the time spent by the pharmacist, pharmacy technician and GP with the average wage of these healthcare providers based on an earlier report presenting costs associated with a CMR [31]. Analysis Descriptive statistics were used to describe patient characteristics. Costs were calculated over the 6month period. To account for missing data in effects and costs (e.g. due to patients not being reachable), the method of multiple imputations was used to generate ten imputed data sets with predictive mean matching, assuming that the data were missing at random. The effectiveness of the intervention was expressed in estimators that are important for patients' daily lives, namely HR-QoL and health related complaints with an impact on patient's daily life. Results of the cost-effectiveness analysis were expressed in terms of the incremental cost effectiveness ratio (ICER) 6 months after the intervention. These ICERs were calculated for all three outcomes: (1) costs/QALY measured with EQ-5D health utility values, (2) costs/QALY measured with EQ-VAS scores, and (3) costs/reduced complaint with impact. The total costs included drug costs, all healthcare costs including informal care and intervention costs, calculated over 6 months from the start date of the study. In order to analyze the uncertainty of the ICER results, we performed a probabilistic sensitivity analysis (PSA) with 1,000 replicationswith gamma distributions for all costs and health-related complaints with impact, a normal distribution for health utility values and a beta distribution for EQ-VAS scores. The resulting 1,000 replicates were plotted on the cost-effectiveness plane and used to construct a cost-effectiveness acceptability curve. The graphical presentation of the cost-effectiveness is presented as the difference in costs on the vertical axis and the difference in effects on the horizontal axis. Deterministic sensitivity analyses (DSA) were conducted for all different cost parameters to test the robustness of the analyses. Estimates for all different types of costs in both groups were varied between their 95% confidence intervals to assess the confidence. The resulting ranges of costs are presented in a tornado plot. Base case analysis shows unadjusted values. An additional analysis, in