Patients were randomly assigned 1 1 to RT versus observation as described by Zelen 15 Fig 1 A using a permuted block random assignment scheme buy cheap cialis online