The study was designed as a double- blind, randomized, placebo-controlled parallel group study. Randomization was performed by allocating patients to consecutively numbered treatment packages. Treatment was randomized by a computer in blocks of four, the block size only being disclosed at the completion of the study.