CN1 Mean Versus Median Overall Survival (OS) For Describing Value of New Cancer Therapies: A Case Study

Abstract

OBJECTIVES:The impact of new oncology therapies on OS is often assessed by comparing median OS times in randomised controlled trials. Although this data is usually available even when many patients remain alive at the end of the trial, the survival times of those surviving beyond the median point may not be adequately accounted for in this comparison. In this case study, we discuss the median and the mean OS using data from a recently published randomised trial. METHODS: Median OS in the ipilimumab-alone (IPI) and gp100 alone-arms of the trial of IPI in pre-treated metastatic melanoma (MM) patients (Hodi et al., 2010, NEJM) was compared with non-parametric estimates of mean survival (area under digitised Kaplan-Meier survivor function) over four years (maximum follow up 55 months). We reviewed the methods literature and approaches adopted in relevant assessments. RESULTS: In this case study, for MM population followed over four years median OS was reached in the control arm at 6.4 months, and at 10.1 months in the IPI alone arm, a difference in medians of 3.7 months. Mean OS (area under the curve) over 4 years was 11.5 months in the control arm and 17.6 months in the IPI alone arm, a difference for IPI of 6.1 months. Though larger than the difference in median OS, this represents a lower bound on the mean OS benefit over the remaining lifetime, since the survival benefit was truncated at the end of the trial. CONCLUSIONS: Mean and median OS both have a place in characterizing OS. In this case study, it would appear that mean OS may be more informative in describing the potential benefit of the treatment in patients with MM. Health care decision makers should consider all the available data when assessing the potential benefits offered by new therapies in oncology. OBJECTIVES:Within cost effectiveness analysis, joint uncertainty in costs and effects is commonly dealt with using probabilistic sensitivity analysis (PSA). Although economic models using patient level data can simulate more complex disease processes than cohort-based models, the computational time required to eliminate 1st-order uncertainty often makes extensive PSA impossible. To overcome this, a non-parametric artificial neural network (ANN) simulation meta-modelling method is presented using a case study that evaluates the cost-effectiveness of intensive blood-glucose monitoring in patients with type 2 diabetes. METHODS: