PCN43 - Forecasting Outpatient Pharmaceutical Expenditure for Cancer Treatment in Germany

Abstract

If a new drug is supposed to replace existing pharmaceuticals we subtracted costs of replaced drugs [table 2]. If a new drug is supposed to be given additionally, we added costs of new drugs. For calculations, we assumed that newly launched drugs will reach on average a diffusion of 20% of their market potential until 2016. Market potential was defined as for a new drug eligible patients. To assess the impact of new drugs, we obtained expert opinion by IMS Health on (a) the timing of drug launches in the German cancer market (b) the expected prices of new drugs (c) the extent to that new drugs will replace existing pharmaceuticals. To forecast the pharmaceutical expenditure until the end of 2016 we used CPE of TK. We incorporated trends in membership to TK and demographic changes. We incorporated costs of new cancer drugs that will be launched until the end of 2016. Finally, to extrapolate results to whole SHI, we adjusted for differences in demographics of insured between TK and SHI using publicly available data, i.e. KM6 statistics. We assumed an increase of average age in SHI and a nearly unchanged number of insured. We performed a Monte Carlo (MC) simulation using the examples of NSCLC and Breast Cancer to consider the uncertainty of our model. We chose NSCLC because we expect the highest number of new drugs for this indication and expect new drugs to replace existing pharmaceuticals as well as to be given additionally [table 2]. We also chose Breast Cancer because we expect the highest pharmaceutical expenditure of the 12 cancer indications in 2016 to be spent for Breast Cancer [table 2]. We approximated the annual expenditure in 2016 based on TK data from April 2012 to March 2013. Therefore we defined probability distributions and generated random values by a sample size of 100,000 for (a) the expected monthly costs of new drugs (b) the extent to that new drugs will replace existing pharmaceuticals (c) the rate of market penetration. We used Statistical Analysis System (SAS) version 9.3 for computation of MC simulation.