As described in the patient exclusion criteria, patients with censored follow-up of less than 1 year

with no evidence of disease progression or death were excluded from the analysis. The primary cost measures included progression-free costs and overall costs. Costs were based on all billed charges documented in the PMS data and were obtained for chemotherapy administration, supportive care, and other costs. Supportive care costs included charges for erythropoietin-stimulating agents, white blood cell growth factors, and bisphosphonates. Other costs included charges for facility costs, physician and nursing fees not included in chemotherapy administration costs, and other ancillary drug and service costs (e.g., laboratory tests and diagnostics). Ku 0059436 Incremental effectiveness was measured as the difference in PFS and OS within each matched LBH589 pair. Incremental costs were measured as differences in costs during the PFS/OS periods for each matched pair. Mean differences in effectiveness and costs were calculated for each matching strata. An overall mean was calculated as the weighted average of the strata-specific differences (weighted by the relative proportional sizes of the strata) in effectiveness and cost. Traditional survival analyses (Kaplan–Meier analysis and Cox proportional hazards regression modeling) were conducted to estimate and compare PFS

and OS between treatment cohorts. These analyses were repeated for those patients who received Pem/Cis as well as for patients with an Eastern Cooperative Oncology Group performance factor (ECOG PS) 0/1 who received Pem/Cis. Statistical inference associated with the mean difference in cost effectiveness and individual

component variables of effectiveness and cost were determined using a bootstrap method, in which the matched pairs in the study population were randomly re-sampled (5000 samples Megestrol Acetate with replacement) [9]. The random re-sampling was done such that, for each sample the same method of estimation for strata mean values and overall mean values of cost effectiveness was repeated as described above for the point estimates. Re-sampling and re-analysis provided an ordered empirical sampling distribution for each mean value, and central 95% confidence intervals (CIs) were constructed using the 2.5th and 97.5th percentiles of each statistic’s empirical sampling distribution. The bootstrapped samples were used to calculate probability for each incremental cost-effectiveness pair falling into any particular quadrant of the incremental cost-effectiveness plane as a proportion of the empirical distribution lying in that quadrant. Statistical analyses were performed using SAS® software, Version 9.1 (SAS Institute Inc., Cary, NC, USA). A total of 481 patients received Pem/Plat therapy during the study period.