Clearly the risk for a drug payment bundle could be prohibitive. The variation from practice to practice, or even doc to doc can be significant. Not all cancers were created equal. Even just in breast cancer the differences between someone who treats hormone positive cancer can be dramatically different from someone who treats Her-2 positive or triple neg patients. As one of the authors pointed out the current ICD10 system makes if very difficult to differentiate these sub categories. Historical utilization and risk stratification could be helpful. If someone treats a higher risk population then their "profile" would reflect that and appropriate adjustments could be made. The usage of disease pathways might also make it easier to predict drug cost and bundling. Finally there should be exceptions for outliers. When the DRG system was first introduced to the hospitals, there was a mechanism to deal with outlier populations. Although there really is significant risk there may be a solution if some of these ideas were considered.