Resource utilization data

were elicited using expert opinion and multiplied with relevant unit costs from appropriate public sources. All costs and outcomes occurring beyond 1 year were discounted at 3%. Results demonstrated higher LY gained for sorafenib compared to BSC associated with higher total cost. The model calculated an overall incremental cost-effectiveness ratio for sorafenib, compared to BSC, of $US62 473/LY gained. Results were sensitive to changes in the discount rate, OS with sorafenib and BSC, and the TTP with sorafenib. The probabilistic sensitivity analysis accentuated the validity of the analysis, and showed that the probability of sorafenib providing a cost-effective alternative to BSC was 68% at $US75 000 Pifithrin-�� and 86% see more at $US100 000.

The results indicate that sorafenib is cost-effective compared to BSC, with cost-effectiveness ratios within the established threshold that society is willing to pay, that is, $US50 000–$US100 000,29 and significantly lower than alternative thresholds that have been suggested in recent years ($US183 000–$US264 000/LY gained) for oncology products.30,31 In the USA, in a survey of 139 academic medical oncologists, the implied cost-effectiveness thresholds, derived from the hypothetical scenarios, averaged around $US300 000.31 In the UK, although £30 000 is the threshold for the National Institute of Health and Clinical Excellence for innovative treatments extending life in disease areas with short life-expectancy (using the end-of-life criteria), ICER such as £54 103 for sunitinib in renal cell carcinoma have also been accepted.32 Data constraints led to certain limitations and, as with most economic models, the analysis was based on multiple data sources and was reliant on certain analytical assumptions. First, in the absence of licensed therapies, a large percentage of this patient population is offered other therapies, such as doxorubicin. These treatment options were not incorporated due to the lack of effectiveness data in this population,

however they would probably increase the cost of the comparator arm significantly, while having limited impact on effectiveness. Thus, incorporating these treatments would further improve the triclocarban cost-effectiveness of sorafenib by decreasing the additional costs associated with the treatment, without any significant change in effectiveness. Second, because the SHARP trial demonstrated a clinically and statistically significant increase in OS at 72 weeks for a sorafenib-treated patient compared to a placebo-treated patient and was consequently stopped early, patient-level data were extrapolated by fitting a distribution to the patient-level data. The results were most sensitive to these efficacy data. For AE, the assumption of being constant throughout the model was made. In clinical practice, these AE are likely to occur in earlier stages of treatment as seen in clinical trials of sorafenib in patients with advanced renal-cell carcinoma.